A media component focusing on experiences associated with different stages of smoking was followed by a structural discussion that linked cognition about symptoms with skills to resist influences to smoke. The control group viewed three commonly used smoking prevention films, made written comments on the films, and participated in an unstructured discussion. Students exposed to the experimental program showed superior gains in accurate interpretations of symptoms, adaptation to symptoms, and addiction. Significantly fewer students exposed to the experimental program made the transition from a first try to occasional or regular smoking at an 18-month followup. Change in understanding of symptoms mediated treatment effects. Obtaining these results with only a three-session program, using an active treatment control, a within-school design, and a racially balanced urban population supports the inclusion of a cognitive-developmental component in smoking prevention. 4 tables, 32 references. (Author abstract modified)