Effects of problem-based learning as reported in curricular comparison studies have been shown to be inconsistent over different medical schools. Therefore, we decided to summarize effects of a single well-established problem-based curriculum rather than to add up sometimes-conflicting findings from different problem-based curricula. Effect sizes were computed for 270 comparisons. The results suggest that students and graduates from the particular curriculum perform much better in the area of interpersonal skills, and with regard to practical medical skills. In addition, they consistently rate the quality of the curriculum as higher. Moreover, fewer students drop out, and those surviving need less time to graduate. Differences with respect to medical knowledge and diagnostic reasoning were on average positive but small. These outcomes are at variance with expectations voiced in recent contributions to the literature. They demonstrate that constructivist curricula can have positive effects on learning even if they deemphasize direct instruction. (Contains 1 table.)