OBJECTIVE To evaluate the incidence and preventability of adverse drug events (ADEs) and to determine the yield of several strategies for identifying them. METHODS Prospective cohort study. METHODS Seven units, including two medical, two surgical, and two obstetric general care units and a coronary intensive care unit in an urban tertiary care hospital. METHODS All patients on these units over a 37-day period (2,967 patient-days). METHODS None. METHODS Events were identified in three ways: 1) logs were placed on each unit and satellite pharmacy for nurses and pharmacists to record incidents; 2) a research nurse solicited reports of incidents twice daily on each unit; and 3) the nurse reviewed all charts at least daily. Incidents were classified by two independent reviewers as ADEs or potential ADEs. RESULTS The rate of drug-related incidents was 73 in 2,967 patient-days; 27 incidents were judged ADEs, 34 potential ADEs, and 12 problem orders. Fifty different drugs were involved. Physicians were primarily responsible for 72% of the incidents, with the remainder divided evenly between nursing, pharmacy, and clerical personnel. Of the 27 ADEs, five were life-threatening, nine were serious, and 13 were significant. Fifteen (56%) of the 27 were judged definitely or probably preventable. Incidents were discovered about equally often from the logs and by chart review. However, when the incidents in which an ADE was present were compared with the remainder of incidents, the authors found that 67% (18 of 27) of the ADEs were identified only by chart review (p < 0.001), and physicians were more often judged responsible than other personnel (p < 0.001). CONCLUSIONS The authors conclude that ADEs are not infrequent, often preventable, and usually caused by physician decisions. In this study, solicited reporting by nurses and pharmacists was inferior to chart review for identifying ADEs, but was effective for identifying potential ADEs. Optimal prevention strategies should cover many types of drugs and target physicians' ordering practices.