The study examined the association of a school-based tobacco-control program with students’ smoking behaviour over time using three cross-sectional, provincial census datasets (grade 10 students in 1999, grade 11 students in 2000, grade 12 students in 2001). Data were collected from all secondary schools in Prince Edward Island (Canada) using the Tobacco module of the School Health Action, Planning and Evaluation System (SHAPES). The proportion of regular smokers increased from grade 10 (22.3%) to grade 12 (27.8%, χ² = 10.35, df = 1, p < 0.001). Being exposed to different school-based tobacco programs and policies in grades 10 and 11 was not associated with the smoking behaviour of grade 12 students. The strongest predictors of smoking behaviour were having friends or close family members who smoke. This preliminary evidence suggests that programs and policies associated with banning smoking and enforcing smoking restrictions at school may be insufficient unless they also address the influence of smoking peers and family members and link to comprehensive programming within the broader context of other community and policy level interventions.