Using methodology similar to the Harvard Medical Practice study, the authors analyzed the medical records of 1006 hospital admissions for evidence of patient safety incidents and compared these results to data from the hospital's voluntary incident reporting system. Nearly one-quarter of all patients experienced a safety incident, but virtually all of them were detected via medical record review, and few through incident report review. This finding corroborates prior research in this area and suggests that incident reporting systems are unlikely to provide an accurate assessment of the epidemiology of medical errors.