Does peer-tutoring affect students' educational outcomes in problem-based learning? Students' characteristics and outcomes were compared along 14 successive classes of a problem-based learning course in the University of Brasilia medical programme. In the first stage of this time series, 26 teacher-tutored groups were formed; in the second stage, 50 groups had both teacher- and peer-tutoring. Both groups had equivalent characteristics in stages one and two as regards membership size and composition (students' learning style, self-confidence and motivation to learn). Results showed that scores for problem-solving tests and self-evaluation of skills were not significantly different between first and second stage groups. However, scores of meaningfulness of course experience and group work usefulness were significantly higher in the peer-tutoring stage. Significant positive correlations were found between scores of meaningfulness of course experience and both self-evaluation and group work usefulness but not between the first measure and group size or motivation to learn. The findings suggest that peer-tutoring can facilitate group work without sustained loss of cognitive achievement in long range conditions of problem-based learning experience.