In medicine, the development of expertise requires the recognition of one's capabilities and limitations. This study aimed to verify the accuracy of self-assessment for the performance of a surgical task, and to determine whether self-assessment may be improved through self-observation or exposure to relevant standards of performance. Twenty-six senior surgical residents were videotaped performing a laparoscopic Nissen fundoplication in a pig. Experts rated the videos using two scoring systems. Subjects evaluated their performances after performance of the Nissen, after self-observation of their videotaped performance, and after review of four videotaped "benchmark" performances. Expert interrater reliability was 0.66 (intraclass correlation coefficient). The correlation between experts' and residents' self-evaluations was initially moderate (r = 0.50, P <0.01), increasing significantly after the residents reviewed their own videotaped performance to r = 0.63 (Deltar = 0.13, P <0.01), yet did not change after review of the benchmarks. Self-observation of videotaped performance improved the residents' ability to self-evaluate.