One of the first studies to examine the link between quality of care and hospital deaths, this article discusses a novel methodology for investigating the prevalence of preventable deaths. The study involved four phases, starting with a sampling method in which 182 deaths from 12 hospitals were identified for chart review. Clinical experts then prepared discharge summaries, including an opinion on whether a death was preventable, while investigators collected demographic and illness severity information. The final phase called for developing a screening tool to identify patients at admission who were at high risk for dying from preventable causes. Based on their described process, the authors reported that 14% to 27% of deaths might have been prevented. The findings are focused on diagnoses of myocardial infarction, cerebrovascular accident, and pneumonia. The authors suggest that a few conditions accounted for most hospital deaths and that systematic reviews of deaths may serve as catalysts for improved performance.