Glass' et al. (1981) meta-analysis techniques provided a systematic approach for the accumulation, quantification and integration of the numerous research findings. Five major modalities were identified and their effect sizes computed for five distinctly different outcomes: knowledge, attitudes, use, skills and behavior measures. The magnitude of the effect size was found dependent on the outcome measure employed and the rigor of the experimental design. These factors were controlled for through use of a standard regression analysis. Peer programs were found to show a definite superiority for the magnitude of the effect size obtained on all outcome measures. On the ultimate criteria of drug use, Peer programs were significantly different than the combined results of all the remaining programs, Peer programs maintained high effect size for alcohol, soft drugs and hard drugs, as well as for cigarette use. Recommendations are made concerning the effectiveness of the underlying theoretical assumption for the different program modalities. Future programming implications are discussed as Peer programs were identified as effective for the average school-based adolescent population, but the alternative programs were shown to be highly successful for the 'at risk' adolescents such as drug abusers, juvenile delinquents or students having school problems. (Author abstract)