A meta-analysis of randomized, controlled trials of social skills training for schizophrenia was conducted. Outcome measures from 22 studies including 1,521 clients were categorized according to a proximal-distal continuum in relation to the presumed site of action of skills training interventions, with content mastery tests and performance-based measures of skills assumed to be most proximal, community functioning and negative symptoms intermediate, and general symptoms and relapse most distal. Results reveal a large weighted mean effect size for content-mastery exams (d = 1.20), a moderate mean effect size for performance-based measures of social and daily living skills (d = 0.52), moderate mean effect sizes for community functioning (d = 0.52) and negative symptoms (d = 0.40), and small mean effect sizes for other symptoms (d = 0.15) and relapse (d = 0.23). These results support the efficacy of social skills training for improving psychosocial functioning in schizophrenia. (Contains 1 figure and 2 tables.)