The survey used a stratified single-stage cluster design with paired selection of cluster replicates. The surveys sampled 3,270 students in 1981, 4,737 students in 1983, and 4,267 students in 1987. Students were selected from more than 200 schools in each survey year. Results revealed that exposure to alcohol education in Ontario schools increased significantly between 1981 and 1987. The proportion reporting drinking in the past year and past 4 weeks declined significantly between 1983 and 1987, although a significant but numerically small increase occurred in the proportion who reported being drunk in the past 4 weeks. The proportion exposed to marijuana education also significantly increased, although the level of exposure was much lower than for alcohol education. The proportion reporting marijuana use in the past year and in the past 4 weeks declined significantly, but the decline was in grades not experiencing increased exposure. However, results among different grade levels were inconsistent. Findings suggested that alcohol education may have an impact on overall levels of drinking, probably on student decisions about whether or not to start drinking. In addition, the impact is on younger students and lighter drinkers. Findings also called into question the effectiveness of increased exposure to marijuana education. Tables and 12 references