Objective: To assess the effectiveness of routine reporting to identify surgical adverse outcomes in comparison with retrospective medical record review. Design: Independent assessment of two methods applied to one sample. Surgeons and surgical residents routinely reported all adverse outcomes for patients in their care during admission. A trained research assistant, blinded to the surgeons' reporting data, retrospectively reviewed the medical records of selected patients and registered all adverse outcomes identified from paper or electronic patient records. Setting: Dutch university hospital. Study sample: A 5% sample of patients (N = 150) discharged in 2002 was taken; oversampling of patients undergoing reoperations, sick patients (ASA ≥3), and those undergoing technically complex surgery was done to increase the yield of adverse outcomes. Main outcome measures: The number of adverse outcomes identified by each method was compared with the total number identified by either method. This was done both for all adverse outcomes and for serious adverse outcomes. Results: Routine reporting identified fewer adverse outcomes than medical record review (62.5% v 78.2%). Complete agreement was achieved in only 40.7% of adverse outcomes. Routine reporting identified slightly more serious adverse outcomes (84.8% v 79.5% of the total), but this difference was not statistically significant. Extrapolating these results to the total number of admissions in 2002, routine reporting underestimated the annual adverse outcome incidence by 1.8% (increasing from 14.5% to 16.3%) and the incidence of serious adverse outcomes by 0.3% (increasing from 6.9% to 7.2%). Conclusions: Neither method identified all adverse outcomes. Routine reporting underestimated the incidence of minor events but was as accurate as record review in identifying serious adverse outcomes.