The effects of 47 peer-based health education programs described in 36 published studies were examined. The over-all effect size was small: the mean d was .190 when controls received no program and .020 when controls received an alternative program. Programs were divided into those focusing on preventing or reducing smoking and programs on other health issues; the latter were further divided into primary prevention and secondary prevention programs. Differences among studies suggested several biases which were likely to have influenced the effect sizes. Preventive interventions that produce only small effects can be valuable because many participants would not have developed the problem even without the program. This review suggested that, when health education programs are studied, (a) detailed statistical information should be provided to facilitate using the research findings in meta-analyses and (b) the costs of innovative programs should be presented to judge whether the results are worth the cost.