<p>Abstract</p> <p>Background</p> <p>Problem-based Learning (PBL) has been suggested as a key educational method of knowledge acquisition to improve medical education. We sought to evaluate the differences in medical school education between graduates from PBL-based and conventional curricula and to what extent these curricula fit job requirements.</p> <p>Methods</p> <p>Graduates from all German medical schools who graduated between 1996 and 2002 were eligible for this study. Graduates self-assessed nine competencies as required at their day-to-day work and as taught in medical school on a 6-point Likert scale. Results were compared between graduates from a PBL-based curriculum (University Witten/Herdecke) and conventional curricula.</p> <p>Results</p> <p>Three schools were excluded because of low response rates. Baseline demographics between graduates of the PBL-based curriculum (n = 101, 49% female) and the conventional curricula (n = 4720, 49% female) were similar. No major differences were observed regarding job requirements with priorities for "Independent learning/working" and "Practical medical skills". All competencies were rated to be better taught in PBL-based curriculum compared to the conventional curricula (all p < 0.001), except for "Medical knowledge" and "Research competence". Comparing competencies required at work and taught in medical school, PBL was associated with benefits in "Interdisciplinary thinking" (Δ + 0.88), "Independent learning/working" (Δ + 0.57), "Psycho-social competence" (Δ + 0.56), "Teamwork" (Δ + 0.39) and "Problem-solving skills" (Δ + 0.36), whereas "Research competence" (Δ - 1.23) and "Business competence" (Δ - 1.44) in the PBL-based curriculum needed improvement.</p> <p>Conclusion</p> <p>Among medical graduates in Germany, PBL demonstrated benefits with regard to competencies which were highly required in the job of physicians. Research and business competence deserve closer attention in future curricular development.</p>