Objectives: To evaluate the impact of school-based mentoring for adolescents (11-18 years) on academic performance, attendance, attitudes, behavior, and self-esteem. Method: A systematic review and meta-analysis. The authors searched 12 databases from 1980 to 2011. Eight studies with 6,072 participants were included, 6 were included in meta-analysis. Studies were assessed using the Cochrane Collaboration Risk of Bias Tool. Results: Across outcomes, effect sizes were very small (random effects), and most were not significant. The magnitude of the largest effect (for self-esteem) was close to zero, g = 0.09, [0.03, 0.14]. Conclusions: The mentoring programs included in this review did not reliably improve any of the included outcomes. Well-designed programs implemented over a longer time might achieve positive results. (Contains 3 tables and 7 figures.)