Doppler echocardiography is often used in evaluating patients with chest pain, but information on prognostic value of this testing and data to help guide selective use are limited. We prospectively studied 448 patients admitted from the emergency department for acute chest pain to assess the utility of qualitative echocardiographic data in predicting long-term survival and the incremental value of this information over routine clinical and electrocardiographic data. Doppler echocardiograms, recorded an average of 21 hours after presentation, were analyzed independently by 2 echocardiographers for global left and right ventricular function and valvular disease. Regional function was assessed by wall motion index. Data on long-term survival were collected with an average follow-up of 35.0 +/- 12.1 months. In univariate Cox regression analysis, left ventricular function and size, wall motion index, right ventricular function, and aortic, mitral, and tricuspid insufficiency were significant predictors of total and cardiovascular mortality. In multivariate analysis, moderate or severe left ventricular dysfunction (mortality rate ratio 3.2, 95% confidence intervals 1.8 to 5.8] and more than mild valvular regurgitation (mortality rate ratio 2.0, 95% confidence interval 1.1 to 3.6) were independent predictors of mortality in a model adjusted for clinical and electrocardiographic data. These factors were more common in patients aged >60 years, in those with prior acute myocardial infarction or angina, and in those with rales on physical examination. In the absence of these clinical characteristics, only 8 of 124 patients (7%) had moderate or severe left ventricular dysfunction or valvular regurgitation. In patients with moderate or severe regurgitation, a murmur was noted on the admission physical examination in 41 of 69 cases (59%). We conclude that echocardiographic evidence of moderate or severe left ventricular dysfunction or valvular regurgitation identifies a high-risk group for overall and cardiovascular mortality in patients with chest pain, and this evidence may not be detected clinically.