Smith et al. (1980) analyzed 475 psychotherapy studies and concluded that individuals receiving treatment were better off than 80 percent of the untreated control groups. These studies were criticized on methodological grounds, particularly for failing to enable calculation of an index of effect size. To address these methodological issues, 20 published studies cited in the Smith study, from two treatment domains, i.e., the effectiveness of client-centered therapy (N=17) and transactional analysis (N=3), were coded according to type of therapy, allegiance of investigator, diagnosis, estimation methods, comparison treatment, subject solicitation, treatment center, and source of means. Effect size and population effect size were computed according to the methodology in the original study. An analysis of the results showed a lack of homogeneity of effect size estimates. The success of client-centered therapy and transactional analysis appeared to depend on investigator bias, location of treatment, diagnosis, and presence of a comparison treatment. Specifically, these two treatment methods appeared to be most successful when applied to problems that occur in academic settings. Caution is recommended in generalizing the results of Smith's study, which overly represented academic problems and educational counselors compared to those in medical settings. (BL)