Our objection was to find determinants of long-term outcome in routine data collected for differential diagnosis of suspected acute myocardial infarction. Study population consisted of 263 discharged patients who were initially hospitalized for differential diagnosis of suspected acute myocardial infarction between October 1992 and January 1993. Follow-up time for all cause and cardiac mortality was 5 years. The variables studied as predictors of outcome were computerized ECG, peak creatine kinase isoenzyme MB, peak troponin I, radiographic evidence of pulmonary congestion (cardiac decompensation), treatment for hyperlipidemia, hypertension or diabetes, smoking, previous myocardial infarction, age and gender. Total mortality was 32% at 5 years, of which 77% (64/83) was of cardiac origin. Pulmonary congestion in chest X-ray was the most powerful predictor of outcome (RR=3.3, 95% CI=2.0-5.2, P<0.001). In multivariate analysis congestion (RR=3.3, CI=2.0-5.2) was the only independent predictor of 5-year total mortality in addition to age (RR=1.06, CI=1.04-1.08). These two variables together with previous myocardial infarction (RR=1.9, CI=1.2-3.1) and hyperlipidemia (RR=2. 0, CI=1.1-3.5) were independent predictors of cardiac mortality. Radiographic evidence of cardiac decompensation during hospitalization is a strong and independent predictor of long-term outcome in unselected patients with suspected AMI. The predictive power of cardiac markers is confined to patients without pulmonary congestion.