In order to assess the performance of a surgical pathology laboratory in a university hospital, we have established a comprehensive system of quality audit based on peer review. Each month, 2% of cases received are selected at random and assessed retrospectively by two senior pathologists. The system, which uses semi-quantitative scoring, examines diagnostic accuracy, identifies delay at any stage in the production of reports, evaluates the overall quality of the slides, the presentation of the final report, and the accuracy of the SNOMED coding. Each of these parameters is graded as "satisfactory," "borderline," or "unsatisfactory." In 20 of 518 cases (3.9%) analyzed in 18 months, the microscopic report was unsatisfactory; in six of these cases, the error could have affected patient management. Remediable faults were detected in the macroscopic description of specimens and in the speed and accuracy of report typing by the secretarial staff. In 13 of 18 months, greater than 10% of reports were delayed because of the time taken in microscopic reporting by pathologists. Some (but not all) of this delay was attributable to the "checking out" of pathologists in training. We conclude that this audit system has uncovered substantial deficiencies in our departmental performance, some of which could affect the clinical course of patients. These surprising results suggest that a system of peer review should be adopted more widely.