To compare students' self-evaluations of their performances with the evaluations they received from their tutors in a problem-based course. In the Occupational Therapy and Physiotherapy Programme of the McMaster University Faculty of Health Sciences in 1993-94, a study was conducted of the self-evaluation skills of 30 students in five tutorial groups. Repeated-measures analyses of variance, factor analyses, and Pearson correlations were used to examine the student's self-evaluations in comparison with their tutors' evaluations on six consecutive occasions over a 14-week period. Significant (p < .01) increases were found for both students' and tutors' evaluation scores over the six evaluations. A significant interaction was also found, with the tutors' evaluations being initially lower but eventually higher than the students' self-evaluations. Correlations between students' and tutors' evaluations rose in a sawtooth manner from .49 on the first evaluation to .84 by the sixth evaluation. There was a pattern of diminishing oscillations in the interobserver correlations over the six evaluations. When combined with the steady increase in evaluation scores, the pattern of diminishing oscillations in interobserver correlations was interpreted more as evidence of a negotiation process between students and their tutors than as evidence of improvement in self-evaluation skills.