Research-based substance use prevention curricula typically yield small effects when implemented by school teachers under real-world conditions. Using a randomized controlled trial, the authors examined whether expert coaching improves the effectiveness of the All Stars prevention curriculum. Although a positive effect on students' cigarette use was noted, this finding may be attributed to marked baseline differences on this variable across the intervention and control groups. No effects were found on students' alcohol or marijuana use or on any of several variables thought to mediate curriculum effects. The effects of coaching on teachers may not become evident until future years, when they have moved beyond an initial mechanical delivery of the curriculum. (Contains 3 tables.)