Background Therapeutic decision making in critically ill patients requires both prompt and comprehensive analysis of available information. Data derived from randomized clinical trials provide a powerful tool for risk assessment in the setting of acute myocardial infarction (MI); however, timely and appropriate use of existing therapies and resources are the key determinants of outcome among high-risk patients. Methods Demographic, procedural, and outcome data from patients with MI were collected at 1073 U.S. hospitals collaborating in the National Registry of MI (NRMI 2). Patients were classified on hospital arrival as either “low risk” or “high risk” according to a modified Thrombolysis in Myocardial Infarction II Risk Scale based on predetermined demographic, electrocardiographic, and clinical features. Results Among the 170,143 patients enrolled, 115,222 (67.5%) were classified as low risk and 55,521 (32.5%) as high risk for in-hospital death, recurrent ischemia, recurrent MI, congestive heart failure, and stroke. Using a composite unsatisfactory outcome measure, in-hospital adverse events were had by a greater proportion of patients initially classified as high risk compared with those classified as low risk. By multivariate analysis, age >70 years, prior MI, Killip class >I, anterior site of infarction, and the combination of hypotension and tachycardia were independent predictions of poor outcome in patients with or without ST-segment elevation on the presenting electrocardiogram. High-risk patients with ST-segment elevation were treated with thrombolytics (47.5%) or alternative forms of reperfusion therapy (9.3%) within 62 minutes and 226 minutes of hospital arrival, respectively. High-risk patients offered reperfusion therapy were also more likely to receive aspirin, β-blockers (intravenous, oral) and angiotensin-converting enzyme inhibitors within 24 hours of infarction (all p < 0.0001), survive their event (8.4% versus 21.4% p < 0.0001), and leave the hospital sooner than those not reperfused. Conclusions This large registry experience included more than 150,000 nonselected patients with MI and suggests that high-risk patients can be identified on initial hospital presentation. The current use of reperfusion and adjunctive therapies among high-risk patients is suboptimal and may directly influence outcome. Randomized trials designed to test the impact of specific management strategies on outcome according to initial risk classification are warranted. (Am Heart J 1998;135:786-96.)