The results of the study indicate that smoking initiation was predicted by individual-level demographic factors, and by the contextual factor of attending a school with a high prevalence of established smoking, but failed to be predicted by a school smoking ban. The academic performance of students who indicated there was no school smoking ban was found to worsen, but an increasing proportion of the student body indicated that such a rule existed. Lower socioeconomic status was found to be an independent predictor of smoking initiation as well as poorer academic performance. The findings of the study underscore the need to further examine a range of outcomes of school smoking policies including education outcomes such as academic performance and early school leaving, and to monitor the impact of school policies on the whole population as well as groups vulnerable to smoking. Many factors influence youth to start smoking, and to continue smoking rather than to quit. While tobacco-free school environments are now widely recommended and /or implemented as part of comprehensive tobacco control strategies, the international evidence was not compelling that school smoking bans actually decreased the incidence and prevalence of smoking among youth. School smoking policies should be monitored for educational outcomes, and the impact of policies on groups vulnerable to smoking. A school ban against smoking might not only be ineffective, but could actually prove detrimental as prevention programs designed with a risk orientation may have an unintended negative impact on alienated youth who are the highest risk of substance use and school drop out behavior. This multi-level study was based on cross-sectional self-reported anonymous data from 12,990 students who participated in the 2002 Student Drug Use Survey in the Atlantic Provinces. Tables, references