Despite years of primary research on problem-based learning and literature reviews, no systematic effort has been made to analyze the relationship between tutor characteristics and student learning outcomes. In an effort to fill that gap the following meta-analysis coded 223 outcomes from 94 studies with small but positive gains for PBL students (g = 0.24). Specific sub-group analyses indicate randomly controlled designs may be more sensitive to differences that favor PBL students, even while there is no relationship between tutor content expertise and student learning. Perhaps surprisingly, student learning decreases as tutor experience increases. Limitations and future work are discussed within a context of scholarly and practical significance.