We report a cross sectional survey to analyze the effects of medical training on the smoking habits of Tunisian medical students, and their attitudes and knowledge about smoking. Two groups of medical students were studied. One group was 257 first year students at the Medical Faculties of Tunis and Sfax, in 1987, the other 211 final year students at the same Faculties in 1994 and who had been in the first year in 1987. A questionnaire bases on that of the WHO and International Union against Lung Disease for health professionals was administered. It was completed by 95% of the students. Fifty-four % were men and 46% women, and 70.2% lived in an urban area before attending university (table 1). Nonsmokers were defined as those who had never smoked. Exsmokers were those who had formerly smoked but no longer did so. Smokers were divided into those who smoked occasionally and those who smoked daily. The prevalence of smoking was higher among the final year students than the first year students. Combined daily and occasional smoking was 24.1% among first year students and 37.1% among final year students (table 2). The rates among men for daily smoking were 19.2% in the first year and 38.9% in the final year, whereas for women the corresponding rates were 1.8% and 2% (table 3). The prevalence of occasional smoking among men was 17.8% for the first year and 17.7% in the final year. Among women, this behavior increased from 5.5% to 16.8%. Men exsmokers increased from 6.2% to 16.8% and women from 4.6% to 13.4% from the first to the final year. The proportion of first year smokers who reported a serious attempt to stop was 64.8% and that of final year students was 50%. Protected personal health was the most common reason (table 4). In the final year, 94.1% of te students agreed strongly with the view that smoking is harmful to health (table 5). However, there was substantial underestimation of the contribution of tobacco to causing serious diseases including bladder cancer, coronary artery disease, peripheral vascular disease, emphysema and neonatal morality (table 6). The study evidences insufficient awareness of medical students about their responsibility for health education and prevention. There was little interest in preventive action for patients. Only 4.5% of the final year students felt that they were equipped to advise patients about smoking. Similarly, 65.7% would not advise patients to stop smoking if they had no smoking-related symptoms and did not raise the question themselves (table 7). There were major deficiences in knowledge of preventive measures. Only 45.5% of final year students considered that they had adequate knowledge to advise patients about smoking (table 8) and 72.5% thought that they should have received more specific training about counseling (table 9). This work shows that, like in Africa, Asia and Europe, Tunisian medical students have an unsatisfactory knowledge of tobacco and its effects. There were no substantial changes in the students' knowledge of, or attitude to, smoking between the first and final year of training. Simply stimulating the interest of these future doctors in the problem of smoking is insufficient.