Background. Although self-assessment is an essential component of self-directed adult learning, few data exist regarding the ability of medical students to perform this important task. Therefore, the purpose of this study was to evaluate the ability of medical students to perform self-assessment during a third-year surgery clerkship. Methods. Sixty-eight (34 male, 34 female) third-year medical students assessed their progress at the midpoint of an 8-week surgery clerkship using an 11-item, competency-based evaluation. Students compared perceptions of their performance with a faculty member's assessment using the identical evaluation form. Results. Male students tended to overestimate their midclerkship performance compared with faculty formative and summative evaluations (3.31 ± 0.03 vs 3.23 ± 0.03 and 3.28 ± 0.03) although this did not reach statistical significance. Female students significantly underestimated their midclerkship performance compared with faculty formative and summative evaluations (3.06 ± 0.03 vs 3.40 ± 0.03 and 3.45 ± 0.03, P < 0.05 vs faculty evaluations). Preclerkship academic performance (first- and second-year grade point averages and NBME Part 1 scores) was not predictive of student self-assessment. Finally, women statistically outperformed men on the surgery clerkship (86.6% ± 0.75 vs 83.2% ± 1.20, P < 0.05 vs male students). Conclusions. Female students tend to underestimate their midclerkship performance compared with male students on a surgery rotation. Despite lower self-assessment, female students actually outperform male students. Women may underreport their capabilities when compared with men as a result of gender differences in socialization. These gender differences in self-assessment may be important to recognize when faculty provide feedback to students. © 2002 Elsevier Science (USA).