Chapter 8

Driving under the influence of cannabis[1]

Stan Thompson was 18 when he and four other teenagers from Kanata were killed in a horrific car accident near Perth that summer day. A youth was found responsible for the fatal accident and served eight months of a 12-month sentence. Cannabis and alcohol-impaired driving was the cause. … The year following Stan's death, his father, Greg Thompson, went to local high schools to talk about the tragedy. He spoke to students about what went wrong and how the tragedy could have been prevented. … His message was that driving a vehicle and smoking marijuana does and will affect driving abilities. He pleaded with the kids not to do it. … Cannabis is not a benign substance. There is very little in the way of research that allows anyone to determine levels of impairment related to cannabis and driving ability, much less the levels of impairment related to cannabis and alcohol and driving ability. We have seen in the Manitoba survey, over one-half of the kids that are using cannabis do so in cars and during school hours. There is no technical or scientific ability to test for cannabis impairment. We do not have the technology, scientific data or the research. We do not have the proper legislation. Studies done in British Columbia indicate that 12 per cent to 14 per cent of the drivers involved fatal motor vehicle accidents had cannabis in their systems. The Government of Quebec and the insurance board in Quebec are presently doing road surveys where people are voluntarily submitting to urine or blood tests. The findings in these tests are that between 12 per cent and 14 per cent of those drivers has cannabis in their system while driving. [2]

If there is one issue, other than the effects of cannabis use on young people or the effects of substance abuse, that is likely to be of concern to society and governments, then it is certainly the issue of how it affects the ability to drive a vehicle. We are already familiar with the effects of alcohol on driving, and the many accidents involving injuries or deaths to young people. In spite of the decreases in use noted in recent years, it is not difficult to admit that one fatal accident caused by the use of a substance is already one accident too many.

As it happens, after alcohol, cannabis is the most widely used psychoactive substance, particularly among young people in the 16-25 age group. Casual use occurs most often in a festive setting, at weekend parties, often also accompanied by alcohol. People in this age group are also the most likely to have a car accident and are also susceptible to having an accident while impaired.

We have seen that cannabis affects psychomotor skills for up to five hours after use. The psychoactive effects of cannabis are also dependent on the amount used, the concentration of THC and the morphology, experience and expectations of users. But what are the specific effects of cannabis on the ability to drive motor vehicles? What are the effects of alcohol and cannabis combined? And what tools are available to detect the presence of a concentration of THC that is likely to significantly affect the psychomotor skills involved in vehicle operation?

Here again, the witnesses heard by the Committee vary in their interpretation of the study results. Thus, the Canadian Police Association told us:


Driving while intoxicated by drugs impairs judgment and motor coordination. In one study involving aircraft 10 licensed pilots were given one marijuana joint containing 19 milligrams of THC - a relatively small amount. Twenty-four hours after smoking the joint, they were tested in a flight simulator. All 10 of the pilots made errors in landing and one missed the runway completely. [3]


Two weeks later, Dr. John Morgan of the City University of New York Medical School said in reference to the same study:


A California-based scientist named Jerome Yesavage wrote the study. It was done in the early 1980s, I think, and it attracted enormous attention. … Doctor Yesavage's study … was completely uncontrolled. … As you all have heard, it is difficult to control for marijuana use. When Doctor Yesavage was funded by the federal government to repeat the study with the simple controls that others and I had suggested, they were unable to show any impact of marijuana use after four hours in a similar group of people. Therefore, I believe that the truth is that marijuana use will impact airplane and driving simulators and to some degree driving performance for three hours to four hours after use; however there is no sustained impact. Any impact is relatively minor. [4]


Making reference to Robbe’s work, which we will be examining in greater detail in this chapter, Professor Morgan added:


A Dutch scientist who has for years worked on driving experiments found that marijuana using drivers have a little difficulty staying right in the middle of the road. That is most sensitive test. If you smoke marijuana, you tend to weave a little bit more than completely sober people do. That is important, although there have been no studies to show that that amount of weaving had a gross impact on driving ability. 

The Dutch scientist included in his report that the amount of weaving was approximately the same in individuals consuming very small amounts of alcohol, very small doses of bensodiazopenes and very small doses of antihistamines. [5]


On the same day, Professor Kalant of the University of Toronto responded as follows:


Dr. Morgan referred to some experimental studies this morning. A number of studies, reviewed by Dr. Smiley in the report of the World Health Organization Committee on Health Effects of Cannabis, indicate a fair measure of agreement on what the predominant effects on driving are. The lane control, as Dr. Morgan mentioned, is impaired. The person does not steer as accurately. In addition, there was slower starting time and slower braking time. There was decreased visual search. In other words, when you drive, you must monitor for sources of danger to both sides and not just ahead of you. There was decreased monitoring, decreased recognition of danger signals. The effects were synergistic with those of alcohol. The one favourable thing about cannabis compared with alcohol was that there was less aggressiveness in the cannabis smokers than in the drinkers, so they were less likely to pass dangerously or to speed. Nevertheless, driving ability was impaired not just by weaker, poorer steering control, but also by less alertness to unexpected things that might happen and pose a hazard. 

I will not go into the statistics of actual field studies of the involvement of cannabis in driving accidents. However, I would like to say that a number of studies have shown that there has been evidence of cannabis presence in the blood or the urine of people who have been stopped for impaired driving who did not have alcohol present. [6]


As we can see, and as was the case with respect to the effects and consequences on the health of users, there are divergent opinions about the interpretation of studies and their meaning in connection with the specific effects of marijuana on driving.

This chapter is divided into three sections. The first considers the ways of testing for the presence of cannabinoids in the body. The second analyses studies on the known prevalence of impaired driving, in both accident and non-accident contexts. The third and last summarizes what is known about the effects of cannabis on driving based on both laboratory and field studies. As in the other chapters, the Committee will then draw its own conclusions.



Forms of testing


There are five known media for testing the presence of cannabinoids in the organism: blood, urine, saliva, hair and perspiration.

Blood is the most appropriate medium for detecting recent cannabis use because only a blood analysis can distinguish between the active ingredients of cannabis and metabolites that have no psychoactive effects. However, as we have already seen, blood concentrations of D9THC peak 9 minutes after smoking; after 10 minutes only two-thirds of the concentration remains, and it is down to 5 to 10% at the end of an hour; after two hours, it becomes difficult to detect. Thus not all methods are appropriate for testing because of the strong possibility of obtaining false negatives and false positives. The most reliable method, gas chromatography using mass spectrometry for detection, is extremely sensitive and can also estimate the time that has elapsed between the most recent use and the taking of the blood sample.

We saw in Chapter 7 that there was a dose-response relationship: 25 puffs affect cognition more than do 10 puffs, and 10 have more of an effect than 4. But not much data is available on the relationship between concentration and effects on people, and the ability to answer the key road safety question, namely at what concentration can one consider that faculties are impaired? In France, the D9THC level that constitutes testing positive has been set at 1ng/ml[7] for drivers involved in fatal accidents. Another author has come up with a formula that establishes a relationship between D9THC, 11-OH D9THC and D9THC-COOH to determine a cannabis influence factor with a positive threshold of 10ng/ml. An equal concentration of D9THC and COOH suggest use approximately 30 minutes beforehand, and hence a very high probability of psychoactive effects, whereas a higher concentration of COOH than D9THC suggests that use was more than 40 minutes beforehand. However, a concentration of COOH in excess of 40 μg/l would indicate a chronic user, and hence it becomes impossible to determine when the last use occurred. Other research has established that a blood concentration of 10 to 15 ng/ml suggests recent use, without however being able to give an exact figure.[8]

Urine tests are also frequently employed and remain the most appropriate method for rapidly determining whether subjects have been using. On the other hand, traces of cannabis can remain in urine for weeks. Furthermore, the traces that remain are of D9THC-COOH, an inactive metabolite. Consequently, urinalyses are primarily useful for epidemiological measurements of cannabis use, and cannot contribute to information about impaired driving.

The levels of concentration of D9THC-COOH in urine are very high: for someone who smokes a joint a day, the level is between 50 to 500 ng/ml and may reach several thousands ng/ml in heavy users; the currently recommended threshold level for testing positive is 50ng/ml urine. 

Saliva is a very promising option for road safety because it is non intrusive and can indicate recent use with some accuracy. The presence of D9THC in saliva essentially results from the phenomenon of bucco-dental sequestration during inhalation. Concentrations are very high in the few minutes following absorption, varying between 50 and 1,000 ng/ml, but then decline very quickly in the hours that follow, though they remain detectable for an average of four to six hours. The European ROSITA project compared the reliability of samples taken from urine, perspiration and saliva compared to that taken from blood. Saliva is by far the most reliable, showing an exact correlation in 91% of cases. However, the low level of concentration during the period when the psychoactive effects are active means that sensitive analytical methods are essential. There is unfortunately not yet a sufficiently accurate and reliable rapid detection tool that can be used in driving situations. Hence the driving detection tools correctly identified only 18 to 25% of cases and led to many false negatives.[9]

Perspiration is generally considered poor for detection purposes, because of the persistence of D9THC in sweat, and the fact that it is also excreted into sweat in small quantities.

Hair looks very promising because the significant amount of D9THC can determine time since and level of use (low, moderate, high). However, concentrations are only a few ng per mg of hair, which requires highly efficient testing.

The following table, taken from the INSERM report, summarizes the main characteristics of the various biological testing media; where available, we have added the threshold detection level adopted.



Main Characteristics of Biological Testing Media


Primary cannabinoids

Maximum detection period

Useful for

Methodologies available

Threshold for positive test
























THC (active)













Occasional use: 2 to 7 days

Regular use: 7 to 21 days


2 to 10 hours



Highly variable







2 to 10 hours


Identifies use





Identifies recent use



Not useful



Identifying & monitoring regular user


Confirmation, identification, dosage


Yes, many rapid tests




No, no rapid tests


No, no rapid tests








50ng of D9THC-COOH per ml



not determined



not useful



not determined




1ng/ml (France)



In all instances, the handling and transportation of samples and the toxicological dosages are essential to the quality of the analyses.

There is still considerable uncertainty about thresholds that make it possible to affirm that the presence of D9THC would impair the driver. Furthermore, there is still no reliable rapid screening test to identify recent use (urine tests cannot do this). Moreover, other drugs besides alcohol, including many types of prescription medicines, may have an impact on driving. That is why many authors, and a number of witnesses, suggested to us that Canada adopt the Drug Evaluation and Classification Program (DEC) and recognize police officers trained as Drug Recognition Experts. This practice has now been adopted in most U.S. states (at least 34, as well as the District of Columbia), British Columbia, Australia, Norway and Sweden.

The typical scenario for driving under the influence of psychoactive substances other than alcohol is as follows: a vehicle attracts the attention of a police officer, who pulls the vehicle over and questions the driver; if there are reasonable grounds to believe that the driver is intoxicated, a breathalyser test is administered; however, when the test yields a result below the legal limit, the police officer may still not be convinced that the driver is capable of driving, but how is this to be proven? Before, more often than not, the police officer had to release the driver. As we have just seen, there are no equivalents to the breathalyser test for drugs and medicines, and, for cannabis in particular, traces found in urine in no way establish that use was recent. It was in this context that the police officers working for the Los Angeles Police Department developed the Drug Recognition Expert System (DRE) in the early 1980s. Police officers are given specific training in the detection of people driving under the influence of psychoactive substances and in the use of the DEC.

The system allows police officers who have reason to believe that drivers are intoxicated to call on an officer specially trained in drug recognition, who can then evaluate the driver on the basis of a set of systematic and rigorous factors that are recognized as signs of the presence of drugs. The process involves 12 steps:

·                Breath alcohol test: This test will have been conducted by the police officer who stopped the vehicle. The Drug Recognition Expert is only called in when the test is negative.

·                Interview by the arresting officer: The DRE asks the arresting officer a series of conventional questions: in what condition did he or she find the suspect, what he or she had observed, if he or she found drugs in the vehicle, suspect’s statement, etc.

·                Preliminary examination (the first of three pulse measurements): This involves determining whether there are reasonable grounds to suspect the presence of drugs, and hence eliminate the possibility that there is a medical condition. The DRE observes the suspect’s overall condition, and questions the suspect about health, examines the pupils and gaze, and takes the first of three pulse measurements. If the DRE feels that there are no signs, the suspect is released. If the condition is medical, a medical evaluation is requested. However, if drugs are suspected, the examination continues

·                Examination of the eyes: This consists of three tests: horizontal gaze, vertical gaze and convergence. Apparently when under the influence of any drug, it is impossible to have an involuntary jerky movement of the pupils on the vertical axis without first provoking such movements on the horizontal axis. Thus if there are only vertical jerky pupil movements, it is likely a medical condition (e.g. brain damage). If there is horizontal jerkiness, there are likely drugs involved. To determine horizontal movements, the DRE moves a pen or other object horizontally in front of the suspect’s eyes. For vertical movement, the pen is moved from top to bottom. Furthermore, as certain drugs prevent eyes from being able to converge towards the bridge of the nose, the DRE performs a convergence test by placing the pen or object on the person’s nose and asking the suspect to look at it

·                Divided attention psychophysical tests: The tests include balancing, walking, standing on one leg and the finger-to-nose test

·                Vital signs examination: This is the second of three pulse measurements, as well as a measurement of blood pressure and body temperature

·                Dark room examination: This involves examining the pupils under four different lighting conditions: room lighting, darkness, indirect light and direct light

·                Examination of muscle tone: arm movements

·                Examination for injection sites

·                Questions about suspect’s drug use and living habits

·                Opinion: On the basis of all the evidence, the DRE forms an opinion based on a reasonable amount of certainty

·                Toxicological examination: The purpose of this examination is to corroborate the analysis by the DRE officer. The decision concerning prosecution is made only when the analyses are returned.


The system was standardized in the early 1980s with the assistance of the U.S. National Highway Traffic Safety Administration. It was first tested in a laboratory study.[10] In the study, four Drug Recognition Experts evaluated subjects who had received either a placebo or a dose of drugs. Neither the subjects nor the officers knew who had received the drugs. In 95% of cases, the officers correctly identified the subjects who had not been given drugs. In 97% of cases, they correctly identified the subjects who had been given drugs and in 98.7% of cases, they were able to determine which subjects were under the influence of drugs.

A field study was then conducted in 1985, once again with the assistance of the Highway Traffic Safety Administration.[11] In the study, blood samples of 173 drivers arrested for driving under the influence of drugs were analyzed by an independent laboratory. The study showed that the analyses carried out by the Drug Recognition Expert officers correctly predicted the presence of drugs other than alcohol in 94% of cases. In 79% of the cases, the analyses of the officers identifying the presence of a specific drug turned out to be accurate.

The most complete study was carried out in Arizona in 1994. In this study, the files of over 500 persons arrested for driving under the influence of drugs were analyzed, and toxicological analyses were conducted. The study showed that the toxicological analyses corroborated the conclusions of the officers in 83.5% of cases. Similar studies conducted in other states yielded comparable results: 81.3% in Texas, 84.5% in Minnesota, 88.2% in California, 88.2% in Hawaii and 88% in Oregon.

With respect specifically to cannabis, the expected signs listed in the system are generally the following: no horizontal or vertical shaking, but no convergence in gaze, dilated pupils, accelerated pulse and high blood pressure.

In short, given the limits of detection in the field of the influence of cannabis and the results of these studies, it would appear that it would be highly desirable to adopt the DEC and train police officers in drug recognition.


Epidemiological data


According to a number of the witnesses we heard, more than 40% of people whose driving abilities are impaired would drive under the influence of cannabis. Others have said that approximately 12% of accidents causing injury could be attributed to the use of cannabis. What do the studies reveal?

Data on the frequency of driving under the influence of cannabis (whether on its own or together with other substances) are, for obvious reasons, difficult to obtain. First, for drivers involved in an accident, a positive breathalyzer test means most of the time that no other measurements are taken because a blood alcohol level above the legal limit is enough to take legal action. Second, the methods available to detect the presence of THC are intrusive (blood, urine), unlike the breathalyzer, and hence pose specific legal and ethical problems. Other forms of measurements, such as saliva samples, do not, for the time being, allow roadside detection. Lastly, in studies of all drivers, the consent of drivers is required to take a blood or urine sample, thus limiting the possibility of generalizing results. Nevertheless, we will summarize the main points of a number of studies conducted in recent years.


Studies not involving accidents

Two types of studies were conducted: surveys of all drivers selected at random from the flow of traffic at various times of the day and week, and studies where it was presumed that the people were driving under the influence during police checks. The following table, drawn from the various data available from INSERM, summarizes these studies.



Detection and prevalence of cannabis in Europe and Quebec where no accidents are involved[12]

Reference country



Detection method


Prevalence (%)

No presumption of driving under the influence of psychoactive substances


Germany, Kruger et al., 1995


Netherlands, Mathtijssen, 1998


Italy, Zancanner et al., 1995



Dussault et al., 2000


All drivers



Night drivers on weekends


Night drivers on weekends


Highway drivers (representative survey)


Screening: FPIA saliva

Confirmation: CG/SM saliva

Screening: combined saliva, perspiration and urine test

Clinical screening, clinical and toxicological check (blood, urine)



Breathalyzer (alcohol)


2 234

(of 3 027)



(of 402)


1 237



2 281

2 260

5 281












(in progress)

> 0.8 : 0.8


With presumption of driving under the influence of psychoactive substances


Norway, Skurtveit et al., 1996


Denmark, Worm and Steentoft, 1996


United Kingdom, Scottland,  Seymour and

Oliver, 1999










Screening: immunoassay blood;

Confirmation: CG/SM blood

Screening: RIA blood

Confirmation: CG/SM blood

Screening: immunoassay blood;

Confirmation: CG/SM blood


2 529

















In all, it was observed that the detection rates for the presence of cannabis varied between 1% and 5% when there was no presumption of impaired driving. However, the missing data, which likely resulted from refusals to supply a sample, made it impossible to draw clear conclusions. The studies with presumption of driving under the influence of drugs had clearly higher results: between 10 and 26%. These results do not necessarily reveal a much higher prevalence of driving under the influence of psychoactive substances, but rather a higher level of vigilance by the police. Indeed, as we shall see immediately, the prevalence of cannabis detection in fatal accidents is no higher in Norway (7.5%) than in other countries.


Studies where an accident was involved

It is difficult to compare studies between countries because the detection methods, even in an accident context, varies widely from country to country. We wish to note once again that simply finding traces of cannabis in drivers involved in accidents is not necessarily a sign that its use was the cause of the accident. Nor does the absence of any screening result mean that no one was driving under the influence of cannabis.

The following table, adapted from INSERM results, refers to a number of recent studies in Europe, America and Australia.


Prevalence of impaired driving(ID) when there are accidents [13]




Detection method


Prevalence of cannabis (%)


Meulemans et al., 1997



Alvarez et al., 1997


France, Mura et al., 2001



France, Kintz et al., 2000



Italy, Ferrara, 1990



Norway, Christophersen, 1995


United Kingdom, Tunbridge, 2000


Australia, Longo, 2000


Canada, Cimburra, 1990


United States, Logan, 1996

Casualty accidents (2-wheeled and cars)



Fatal accidents with suspected ID


Casualty accidents (control group: patients)


Casualty accidents





Friday night checks


Injuries, non-fatal accidents



Fatal accidents (including 516 drivers)


Injuries (non-fatal accidents)






Screening: urine

Confirmation: urine CG/SM and urine blood comparison


Screening: immunoassay blood

Confirmation: CG/SM blood


No screening

Confirmation: CG/SM blood



Screening: urine

Confirmation: CG/SM urine and blood, saliva and perspiration tests

Screening: EMIT urine




Screening: immunoassay blood

Confirmation: CG/SM blood



Screening: immunoassay urine

Confirmation: CG/SM blood


Screening: immunoassay blood

Confirmation: CG/SM blood


1 879















4 350







1 138



2 500



1 169





6 (urine)

3.6 (blood)




not reliable






13.6 (urine)

9.6 (blood)






















Three of these studies are particularly interesting. The Mura et al. study (2001) shows a significant difference by driver age: among 18-20 year olds, the D9THC was present in 18.6% of drivers, and in 50% of cases it was present alone (without alcohol). An earlier study by Mura (1999) had shown that cannabis was particularly common among young drivers: from 35% to 43% in the under 30 age group, with an even higher prevalence (43%) for the under 20s, whereas past the age of 35, the prevalence drops to 3%.[14]

The study by Kintz et al. (2000) is interesting primarily because it clearly shows that, after alcohol (13.6%) cannabis is the substance most frequently present among drivers involved in accidents (9.6%). This study also shows that in the whole sample, the incidence of cannabis as measured by taking a blood sample (9.6%) is close to the level of driving under the influence of alcohol (10.6%).[15]

Then, Longo’s study is of special interest because of the size and representativeness of the sample and the fact that separate analyses were done of D9THC and D9THC-COOH. The study detected the presence of cannabinoids in 10.8% of drivers: 8% for D9THC-COOH alone and 2.8% for D9THC-COOH and D9THC together, thereby showing a lower percentage of positive tests for D9THC than the other studies. Furthermore, as in the other studies, subjects testing positive to D9THC were younger and more often men.

Closer to home, Mercer and Jeffery examined the toxicological analyses for 227 drivers killed in traffic accidents in British Columbia between October 1999 and September 1991.[16] Samples had been taken during autopsies within 24 hours of death, which according to the authors, may indicate an under-estimation of the presence of alcohol or drugs. Of the 227 people killed, 186 (43%) showed no signs of either alcohol or drugs, 83 (37%) alcohol only, 23 (11%) alcohol and drugs, and 21 drugs only. As for cannabis, 29 of the people killed (13%; 26 men and 3 women) tested positive to D9THC-COOH, showing an average concentration of 15.9 ng/ml. In the +alcohol/+drugs group, (23 subjects), 17 tested positive to THC metabolites and 8 were also positive to D9THC (13%). For the 0alcohol/+drugs group, (21 subjects), 8 (all men) were positive to D9THC–COOH, and 4 to D9THC. Even though the authors concluded that D9THC /D9THC-COOH was present in 13% of cases, which is a percentage comparable to most of the other studies, only 12 subjects killed tested positive to D9THC with or without alcohol and only 4 without alcohol.

Lastly, a more recent epidemiological study dealt with 1,158 cases of fatal accidents (391) or of cases of driving under the influence of psychoactive substances when the percentage of alcohol in the blood was below 0.1 (767) reported in Canadian forensic laboratories on November 12, 1994.[17] The most frequent substances identified were benzodiazepines (590 cases), alcohol (580), cannabis (551), stimulants (224), opiates (176) and barbiturates (131). For cannabis, we get the following table:




Presence of cannabis in Canada (1994)



with alcohol

without alcohol


Ø Impaired driving


Ø Death



Ø Impaired driving


Ø Death































In all, cases in which D9THC without alcohol was present accounted for 13% of the total, which is close to the figure found in the other studies.

Out of all the studies, it was found that the presence of cannabis among drivers who were injured or killed varies between 3.6% (confirmed by blood analysis) and 13% (unconfirmed). Where there was confirmation of the presence of D9THC compared to D9THC-COOH, the presence of the active substance decreases by half. In addition, the risk of testing positive is much higher for young men than other drivers. These conclusions are largely shared by other authors.[18]


Epidemiological studies on youth

In recent years, epidemiological studies on youth in the school environment have asked questions about the frequency of driving under the influence of psychoactive substances, cannabis in particular. In Ontario, the 2002 OSDUS study described in Chapter 6 shows that 19.3% of the students had driven their car one hour or less after having taken cannabis at least once in the past twelve months.[19] More interesting is that this compares with 15% who said they had taken their car less than an hour after one or two drinks. In Manitoba, the survey of youths in school reveals that almost 20% see nothing wrong in driving after taking cannabis.[20]

Finally, the Cohen and Kaal study on long term consumers had shown that no less than 42% of the respondents in Amsterdam and 74% in San Francisco had driven their car under the influence of cannabis.[21]


Risk assessment

Given the difficulties of conducting reliable epidemiological studies on driving under the influence of cannabis, a number of authors have analyzed the probability of responsibility and the risk ratio involved in the use of cannabis. These studies distinguish between drivers who are responsible for accidents and those who are not. The former are the subjects and the latter the control group. Comparisons are then made of their intoxication to various substances. Clearly, placing drivers into the two categories of responsible/not responsible may depend on an investigator’s perception of whether or not psychoactive substances are present.

The following table, which is reproduced from the Ramaekers et al. report (2002) for the International Scientific Conference on Cannabis summarizes the results of various studies.[22] It should be pointed out that the probability of responsibility for drivers showing traces of cannabis (D9THC and/or D9THC–COOH, whether measured in blood or urine) is compared to the responsibility of drivers involved in an accident not testing positive to any substance (including alcohol). The risk ratio for drivers not testing positive to any substances is 1.0 and is used as a point of comparison to determine the statistical significance of observed change in the risk level of impaired drivers. When the reference value is above the statistical confidence level of 95%, the obvious conclusion is that the drug is 95% associated with an increased risk of responsibility.


Level of culpability relative to driving under the influence of cannabis



Odds ratio

Confidence interval at 95%

N of drivers culpable / not culpable


Terhune & Fell (1982), U.S.




Williams et al. (1985), U.S.




Terhune et al. (1992), U.S.




Drummer (1994), Australia





Hunter et al. (1998), Australia














Lowenstein & Koziol-McLain (2001), U.S.



Drummer et al. (2001) & Swann (2000), Australia



Drug free cases





Drug free cases



Alcohol/THC or THC-COOH


Drug free cases





Drug free cases






Drug free cases



Ø £ 1.0 ng/ml

Ø 1,1 – 2,0 ng/ml

Ø > 2 ng/ml



Ø 1 – 10 ng/ml

Ø 11 – 20 ng/ml

Ø 21 – 30 ng/ml

Ø > 30 ng/ml




No substance





No substance



THC > 5 ng/ml




















































2.8 – 10.5

0.7 – 6.6




2.1 – 12.2

0.2 – 1.5

3.1 – 26.9



5.1 – 10.7

0.2 – 1.8

2.1 – 72.1



3.2 – 9.6

0.4 – 1.5

1.9 – 20.3



4.3 – 11.1


0.3 – 2.1

0.2 – 1.4

0.6 – 5.7



0.5 – 2.2

0.4 – 2.1

0.6 – 4.8

0.6 – 4.8


4.6 – 36.7



1.1 – 9.4

0.5 – 2.4

1.2 – 11.4



4.1 – 8.2

1.2 – 7.6

1.3 – 115.7

0 – 1.3

2.6 – 136.1



















































The study findings show that cannabis alone does not increase the likelihood of responsibility in an accident. However, most of the studies used a measurement of THC-COOH, an inactive metabolite that can remain in urine for several days. When the authors separated out THC alone, the risk ratio was slightly higher, even though it did not reach the required level of significance. In addition, as the concentration of THC increases, the more the ratio increases, once again suggesting a dose-response relationship. Furthermore, the cannabis and alcohol combination significantly increases risk. Without being able to draw any definite conclusions, there are some signs that their effects are in synergy and not merely additive.

Studies on injured drivers (Terhune (1982) and Hunter (1998)) have ratios somewhat higher than in the other studies on fatal accidents. According to Bates and Blakely (1999), the apparent reduction in the risk of a fatal accident stems from the fact that drivers under the influence of cannabis drive less dangerously, for example by reducing their speed.[23]

To conclude, we are rather in agreement with INSERM concerning these studies:


[translation] The findings definitely confirm the significant risk of alcohol, but generally fail to demonstrate that there is an effect of cannabis alone on the risk of being responsible for a fatal accident or an accident involving serious injury. The methodological difficulties that make such a demonstration difficult contribute considerably to the absence of statistically indisputable results. Analyses of responsibility nevertheless suggest that the association between alcohol and cannabis increases the risk of being responsible for an accident, compared to drinking alone; however, this finding needs to be consolidated. Lastly, the most recent data tend to show that there is a risk of becoming responsible at heavy concentrations of D9THC. This involves using cannabis immediately before driving, and perhaps applies also to chronic users. [24]



Experimental studies


Epidemiological studies indicate a relatively high level of driving under the influence of cannabis, between 5% to 12% of drivers, mostly among young men. At the same time, neither these studies nor the responsibility/risk analyses reach clear conclusions concerning the role of cannabis in dangerous driving. Hence the interest in studies on how cannabis affects driving ability and driving itself. Studies on the psychomotor and cognitive skills needed to drive vehicles have measured factors such as: motor coordination, reaction time, attention, visual attention and deductive reasoning. There are two types of studies on driving: simulated studies and field studies, whether on a track, in the city or on a highway. Most studies focus on single doses for recreational users. They use control group protocols and cross-linked protocols, including placebos and comparisons with alcohol. However, they are limited by the fact that they mainly measure the acute effects of single doses, making it difficult to determine whether more experienced users would react in the same way. The following sections examine both types of study.


Non-driving activities

In 1985, Moskowitz published a remarkable synthesis of studies on the psychomotor and cognitive effects of cannabis.[25] In this synthesis, he examined motor coordination, reaction time, tracking and sensory functions. The author observed the following:

·                motor coordination, measured by hand stability, body balance and movement accuracy was significantly affected. However, the application of these results to driving a car is limited, except in driving situations that require considerable coordination, such as emergency situations. The limits in terms of dose and number of subjects tested (between 8 and 16) also need to be noted

·                reaction time was not significantly changed: “There are a sufficient number of experiments involving both simple and complex reaction time situations to leave us relatively well assured that neither the speed of initial detection nor the speed of responding are, per se, impaired by marihuana. Rather, when marihuana produces a reaction time increase, there is some dimension of the information processing task which the subject must execute which bears the brunt of the experiment.”[26] Attention rather than reaction time was affected by marijuana use

·                straight line: this dimension was particularly sensitive to the effects of marijuana, and the vast majority of studies showed a significant reduction in the ability to go in a straight line or correct deviations from the line

·                the sensory functions (hearing and visual) are often affected, but the studies did not yield precise results concerning the distinction between simple tasks and complex tasks.


Ramaekers et al. (2002), reported a meta-analysis on 87 controlled laboratory studies on the psychomotor effects of cannabis conducted by Berghaus et al. (1998). These authors found that the number of psychomotor functions linked to driving (following, reaction time, perception, hand-eye coordination, body balance, signal detection and divided and continuous attention) affected by THC reached a maximum during the first hour after smoking, and one to two hours after oral ingestion. The maximum figures were comparable to those obtained with an alcohol concentration equivalent to > 0.05 g/dl. The number of functions affected reached zero after three to four hours, and only higher doses continued to have an effect. The studies surveyed also showed that THC concentration in the blood is highly correlated to psychomotor effects: a concentration of between 14 ng/ml and 60 ng/ml affected between 70% and 80% of tasks.[27]

The following table summarizes these data:



Deterioration of performance on psychomotor tests by dose,

time and method of ingestion

THC dose

Time (in hours)

        < 1                     1-2                      2-3                      3-4                        4-5              


Tests (n)  % affected

Tests (n)  % affected

Tests (n)  % affected

Tests (n)  % affected

Tests (n)  % affected


< 9mg

9 – 18 mg

³ 18 mg




< 9mg

9 – 18 mg

³ 18 mg




271        61%

193             53%

64                 64%

528             58%



3           33%

 3             0%

3                        0%

9            11%


33                     36%

 48           38%

 28           36%

109                   37%



 49           14%

 41           39%

 45           60%

135          37%


 10           30%

   8           38%

 10           40%

28                     36%



 37            8%

 45          18%

 15          33%

 97          20%


 10            0%

   6            0%

 15          53%

 31          26%



 13            8%

 17           18%

 15           33%

 45           20%


 11            0%

   2            0%

   3           67%

 16           13%



  -                -

  -                -

 11           45%

 11           45%



More recently, after surveying the studies carried out in recent years, the reports prepared by INSERM and the International Scientific Conference on Cannabis reached largely similar conclusions: cannabis affects reaction time where choice is involved, road tracking, shared attention and continuous attention, as well as memory processes, but does not significantly affect simple reaction time or visual or eye-movement functions.


While driving

One of the weaknesses of the laboratory studies is the difficulty of relating psychomotor and cognitive tasks directly to driving. Several tests measured in these studies are short and relatively simple and do not necessarily reflect real situations. The advantage of simulated driving studies and field driving studies is that it brings the conditions closer to reality.

Most contemporary studies have similar characteristics: subjects have had a driver’s licence for at least three years. They are often regular cannabis users. The subjects receive either cannabis or a placebo in a double-blind situation that is very strictly timed to control the level of THC transmitted. In some instances, the experimenters also include comparisons with alcohol and an alcohol placebo. However, it is impossible to control how much subjects inhale and actually absorb. The cannabis prepared by the U.S. National Institute of Drug Abuse (NIDA) varies between 1.75% THC for low doses, 2.67% for moderate doses and 3.95% for strong doses. Converted into mg/kg of weight, the doses correspond to 100, 200 and 300 mg/kg, whereas the heavy dose usually preferred by regular users is generally 308 mg/kg. The subjects are familiarized with the equipment used and the tasks to be performed, and are accompanied by instructors on actual driving studies. Measurements include the standard deviation of lateral position in relation to the road, the control over longitudinal position (distance) in relation to the vehicle ahead, decision-making in emergencies, style of driving and risk taking.

The following table, adapted from INSERM data, summarizes a number of the more recent studies.



Effects of cannabis on car driving[28]

Reference / environment


Subject / Dose / Protocol





Liguori et al., 1998














Sexton et al., 2000






10 users


Cigarette 1.77% THC smoked in 5 mn

Cigarette 3.95% THC smoked in 5 mn

Test: 2 mn after

Duration: 1 hour








15 users


Grass, low dose 1.77% THC

Heavy dose: 2.67% THC

1 resin cigarette: 1.70% THC

Blood and saliva sample 10 mm after start

Test 30 mn

Duration: 25 mn







Avoid a barrier that suddenly appears by braking (55 to 60mph)







Judgment: maintain speed of 30mph on marked road and select widest lane at intersection


Highway section with vehicle ahead passing



Highway section with vehicle ahead braking


16.7 km of highway section



Left and right turns





Intersection with traffic lights, with 4 lane road








Total braking time




Lag time to take foot off accelerator and step on brake



Average speed Number of cones knocked over

Number of successful choices


Average reaction time



Average reaction time


Maximum, minimum and average speed


Standard deviation for perfect line




Response time in going through amber


Average waiting period at a point 10m from the stop line


↑ Slightly significant at 1.77 THC, slightly more at 3.95


No difference




No effect






↑ At low dose (high level of variability at heavy dose: ns)


↑ At low dose (ns)



↓ Average of 6mph at low and heavy dose


↑ Variation at heavy dose versus low dose or placebo


↓ At heavy dose




↓ At heavy dose (high level of variability: ns)

Actual driving

Robbe, 1998 study No. 1 Closed portion of highway (cannabis)





Study No. 2

Normal traffic on highway (cannabis)








Study No. 3

City driving (cannabis)






Study No. 3

City driving (alcohol)






Robbe, 1998

Highway driving (cannabis and alcohol)
















Lamers and Ramaekers, 2000

City driving (cannabis and alcohol)



24 users


100, 200 and 300

Test: 40 mm and 1 hour 40 mm after





16 users

same doses as study 1

Test: 45 mn after









16 users



Test: 30 mn after





16 users


Alcohol level: 0.5 g/l






18 users

THC: 100, 200

Alcohol: 0.4 g/l


Alcohol 0 + THC 0

Alcohol ) + THC 100

Alcohol 0 + THC 200

Alcohol 0.4 + THC 0

Alcohol 0.4 + THC 100

Alcohol 0.4 + THC 200

Alcohol plus cannabis 60 mn after

Tests between 9:00 p.m. and 11:15 p.m.






16 users

THC 100

Alcohol 0.5 g/l

4 preparations:

Alcohol 0 + THC 0

Alcohol 0.5 + THC 0

Alcohol 0 + THC 100

Alcohol 0.5 + THC 100

Tests: 15 mn after

Duration: 45 mm


Constant speed at 90km/hr and tracking over 22km







Tracking control (Ibid.) 64km, 50 mn


Following cars over 50m at variable speed (between 80 and 100km/h) over 16 km, 15 mn




City driving 17.5 km

Dense, moderate or light traffic














Tracking: speed at 100km and constant lateral position








Following: follow a vehicle over 50 m with speed varying by ± 15km/hr every 5mn


Driving in traffic




City driving 15 km




Visual search monitoring


Standard deviation of lateral position


Average lateral position deviation


Average speed and standard deviation


Same measurements


Average reaction time


Average distances and standard deviations



External observations


Internal observations: skill, manoeuvres, turns…


External observations


Internal observations: skills, manoeuvres, turns…


Standard deviation of lateral position









Reaction time





Average distances and standard deviations


Frequency of appropriate eye movements


Quality of driving







↑ Instability at all 3 doses


No effect



No effect



Same effects



↑ ns



Distance increased by 8, 6 and 2 m for 100, 200 and 300 THC


No significant change


No effect





No significant change


0.34 g/l alcohol level modifies control and manoeuvres


↑ Tracking variability; low alcohol alone, THC 100 alone; Moderate: THC 200

Heavy: alcohol 0.4 and THC two doses


↑ Reaction time for 0.4 alcohol and THC 200



↑ Variability in distance between cars in all cases


No effect with alcohol alone or cannabis alone


↓ Performance if alcohol + cannabis

No effect




It is interesting to recall that one of the first driving studies on the road was conducted for the Le Dain Commission.[29] In this study, on a closed track, 16 subjects were each given the 4 following preparations: placebo, marijuana 21 and marijuana 88 μg/kg THC and a dose of alcohol equivalent to BAC 0.07. The tests were conducted immediately after use and three hours later. The subjects were to complete six circuits of the track (1.8 km) with manoeuvres involving slowing down while going forward and backwards, maintaining a trajectory and weaving through cones. The alcohol and heavy dose of marijuana decreased driver performance in tests conducted immediately after use. At the heavy cannabis dose, drivers drove more slowly. On the second test, the differences were less clear.

When the results of this study are compared to those conducted more recently using much more sophisticated methods, it can be seen that the results are remarkably similar.[30] Thus the following was observed:

·                lateral control: this is the variable that is most sensitive to the effects of THC, but the effects are variable, depending on the dose and time; only heavy doses significantly affected lateral control over the vehicle. In comparison, alcohol has a greater effect on vehicle lateral control and speed (linked variables)

·                speed control: in almost all cases, the use of cannabis significantly decreases speed

·                risk-taking: in addition to decreasing speed, it is generally found that there is an increase in distance between vehicles among marijuana users, and less of a tendency to pass or attempt dangerous manoeuvres

·                decision time: this variable is particularly important in actual driving situations. The results do not appear to be very consistent. Smiley suggests that reaction time is unaffected when the subjects are told that they need to respond rapidly, whereas on the other hand, when the obstacles are completely unexpected, the subjects who used cannabis do not perform as well

·                combined effects of alcohol and cannabis: when the researchers checked the effects of the two substances, the combined effects of cannabis and alcohol were systematically greater than alcohol alone or, even more so, than cannabis alone.


Lastly, with low doses, subjects had the impression that their driving was not as good as observers felt it was, which was not necessarily the case with higher doses, where the perceptions of both the drivers and the observers agreed.





The Committee feels it is likely that cannabis makes users more cautious, partly because they are aware of their deficiencies and they compensate by reducing speed and taking fewer risks. However, because what we are dealing with is no longer the consequences on the users themselves, but the possible consequences of their behaviour on others, the Committee feels that it is important to opt for the greatest possible caution with respect to the issue of driving under the influence of cannabis. Given what we have seen in this chapter, we conclude the following.



Conclusions of Chapter 8

Epidemiological data








Data on effects on driving














Further studies



Ø            Between 5% and 12% of drivers may drive under the influence of cannabis; this percentage increases to over 20% for young men under 25 years of age.

Ø      This in itself does not mean that drivers under the influence of cannabis represent a traffic safety risk.

Ø      A not insignificant percentage of drivers test positive for cannabis and alcohol together.


Ø            Cannabis alone, particularly in low doses, has little effect on the skills involved in automobile driving.

Ø            Cannabis, particularly in the doses that match typical doses for regular users, has a negative impact on decision time and trajectory.

Ø            Cannabis leads to a more cautious style of driving.

Ø            The effects of cannabis when combined with alcohol are more significant than for alcohol alone.


Ø            Blood remains the best medium for detecting the presence of cannabinoids.

Ø            Urine cannot screen for recent use.

Ø            Saliva is promising, but rapid commercial tests are not yet reliable enough.

Ø            The visual recognition method used by police officers has yielded satisfactory results.


It is essential to conduct studies in order to:

Ø            Develop a rapid testing tool.

Ø            Learn more about the driving habits of cannabis users.

[1]  In addition to the specific studies we consulted, which will be referred to appropriately, this chapter is largely based on the surveys carried out by INSERM (2001) op. cit., Ramaekers et al., for the International Science Conference on Cannabis in Pelc, I., op. cit.), and Smiley (1999) in Kalant (ed.) op.cit.

[2]  R.G. Lesser, Chief Superintendent, Royal Canadian Mounted Police, testifying before the Special Senate Committee on Illegal Drugs, October 29, 2001, Issue 8, page 17.

[3]  Dale Orban, Detective Sergeant, Regina Police Service, for the Canadian Police Association, testimony given before the Special Senate Committee on Illegal Drugs, May 28, 2001, Issue 3, page 47.

[4]  Dr John Morgan, Professor at the City University of New York Medical School, testimony before the Special Senate Committee on Illegal Drugs, June 11, 2001, Issue 4, page 40-41.

[5]  Ibid.

[6]  Dr Harold Kalant, Professor Emeritus, University of Toronto, testimony before the Special Senate Committee on Illegal Drugs, June 11, 2001, Issue 4, page 75.

[7]  In this chapter, ng means nanogram (i.e. one billion of one gram) and μg means microgram (one million of one gram)

[8]  INSERM (2001), op. cit., pages 152-153.

[9]  Ramaekers, J.G. et al., 2002 “Performance impairment and risk of motor vehicle crashes after cannabis use” in Pelc, I. (ed.) International Scientific Conference on Cannabis, Brussels, page 81.


[10]  Bigelow, G.E. (1985) Identifying types of drug intoxication; laboratory evaluation of a subject procedure. Cited in Sandler, D. (2000) “Expert and Opinion Testimony of Law Enforcement Officers Regarding Identification of Drug Impaired Drivers.” University of Hawaii Law Review 23 (1), 150-181.

[11]  Compton, P.R. (1986) Field Evaluation of the Los Angeles Police Department Drugs Detection Procedure. Cited in Sandler, op. cit., page 151.

[12]  Table reproduced from INSERM (2001), op. cit., page 175.

[13]  Adapted from INSERM (2001) op. cit., pages 171 and 174.

[14]  See INSERM, (2001), op. cit., page 172.

[15]  Ibid.

[16]  Mercer, W.G. and W.K. Jeffery (1995) “Alcohol, Drugs and Impairment in Fatal Traffic Accidents in British Columbia” Accid. Anal. And Prev., 27 (3), pages 335-343.

[17]  Jeffery, W.K. et al. (1996) “The involvement of drugs in driving in Canada: An update to 1994.”  Can. Soc. Forens. Sci. J., 29 (2), pages 93-98.

[18]  Including the INSERM report (2001), op. cit.; Ramaekers, J.G. et al., (2002) “Performance impairment and risk of motor vehicle crashes after cannabis use” in Pelc, I. (ed.) International Scientific Conference on Cannabis, Brussels.

[19]  Adlaf, E.M. et A. Paglia (2001) Drug Use among Ontario Students 1997-2001. Findings from the OSDUS.  Toronto: Centre for Addiction and Mental Health, page 134.

[20]  Patten, D., et coll., (2000) Substance Use among High School Students in Manitoba. Winnipeg: Addictions Foundation of Manitoba.

[21]  Cohen, P.D.A. et H.L. Kaal (2001) The Irrelevance of Drug Policy. Patterns and careers of experienced cannabis use in the populations of Amsterdam, San Francisco and Bremen. Amsterdam: University of Amsterdam, CEDRO, page 68.

[22]  Ramaekers et al. (2002), op.cit., page73.

[23]  Cited in INSERM (2001), op. cit., page 192.

[24]  INSERM (2001), op. cit., page 194.

[25]  Moskowitz, H., (1985) “Marihuana and Driving.” Accid. Anal. Prev., 17 (4), pages 323-345.

[26]  Ibid., page 330.

[27]  Ramaekers J.G. et al. (2002), op. cit., page 77.

[28]  Table adapted from INSERM (2001) op. cit., pages 183-184.

[29]  See Hansteen, R.W, et al. (1976) “Effects of cannabis and alcohol on automobile driving and psychomotor tracking.” Annals of the New York Academy of Science, 282, pages 240-256.

[30]  See notably the survey of studies and the discussion in Smiley, A., (1999) “Marijuana: On-Road and Driving Simulator Studies” in Kalant, H. et al., (ed) The Health Effects of Cannabis. Toronto: Addiction Research Foundation, pp. 173 passim.

Previous DocumentNext Document
Top of document