It is evident that the smoking of marijuana by human subjects does have a detrimental effect on their driving skills and performance in a restricted driving area, and that this effect is even greater under normal conditions of driving on city streets. The effect of marijuana on driving is not uniform for all subjects, however, but is in fact bidirectional; whether or not a significant decline occurs in driving ability is dependent both on the subject's capacity to compensate and on the dose of marijuana. For those subjects who improved their performance, the explanation may lie in overcompensation and possibly the sedative effect of the drug. Whereas the street portion of this study approximated normal driving conditions, it should be emphasized that the context of the driving experience een on city streets was experimental. the design of this study maximal safeguards in terms of a dual control vehicle and a driver observr; in addition, the subjects were proffessionally screened and, with rare exception, they were emotionally stable. Given the experimental setting and set, the safeguards, and the nature of the study sample, idiosyncratic behaviour that might occure under normal driving condition would be less likely to occur in a study such as this. Other identified factors might lead to more stringent conclussions regarding the effects of marijuana on driving.The first is night driving, which may be more stressful. But an even more important unanswered question is the cumulative effect of alcohol and marijuana on driving (64 percent of the study sample reported alcohol in combination with marihjuana before driving). Third, the doses of marijuana used in this study were within the range of social marijuana usage(1); more heroic doses might be taken before driving. Fourth, the effect of marijuana on reactions and decision during high speed is still another unknown. What are the recommendations that emarge from this study? Driving under the influence of marijuana should be avoided as much as should driving under the influence of alcohol. More investigation is urgently required-and high priority should be given to studies that approximate normal conditions of driving and in which alcohol and marijuana are administered to the same subjects.