The impact of analysis of the severity of illness on the in-hospital mortality was reviewed retrospectively on a cohort of patients admitted to a coronary care unit. Three hundred and eighty-six patients were admitted during the study period, with diagnoses including myocardial infarction (49 percent), unstable angina (23 percent), arrhythmias (11 percent), congestive heart failure (5 percent), and nondiagnostic chest pain (5 percent). Total in-hospital mortality was 13 percent. Severity of illness was measured using the acute physiology and chronic health evaluation (APACHE 2) score. Mortality was found to be influenced by admitting diagnosis (p = 0.01), source of referral (p = 0.03), and APACHE 2 score (nonsurvivors, 16.5 +/- 10.1; survivors, 8.5 +/- 4.8; p less than 0.001). The receiver operating characteristic (ROC) curve for the APACHE 2 confirmed it as a predictor of mortality, with an area under the curve (+/- SE) of 0.75 +/- 0.04 (95 percent confidence limits, 0.67 to 0.83). Logistic regression analysis showed the APACHE 2 score and the diagnosis on admission to be significant multivariate predictors of mortality, and a series of diagnosis-specific coefficients are presented. We conclude that the APACHE 2 score is a useful tool for the overall assessment and management of the CCU, as it is in the multidisciplinary medical-surgical intensive care unit.