To determine how physicians interpret exercise electrocardiography with respect to prognosis after acute myocardial infarction (AMI), 29 cardiologists (all board certified) were presented a case history of a 50-year-old man with an uncomplicated AMI and asked to estimate the patient's risk of dying over the next year, the sensitivity and specificity of exercise electrocardiography with respect to 1-year mortality, and the patient's risk of dying given a positive and a negative test result. Each set of physician estimates did not differ from those derived from a review of the medical literature (difference not significant for each). Risk after the test was also calculated using the Bayes' theorem. Calculated versus estimated risks were compared after a negative (7 +/- 9 vs 11 +/- 11%) and a positive (27 +/- 22 vs 17 +/- 15%, differences not significant) test result. Estimated risks were more accurate for a negative result than for a positive one (89 +/- 10 vs 83 +/- 12%, p less than 0.001). Given a positive test result, 57% of the physicians recommended coronary angiography. However, their estimates of risk (30 +/- 23%) were not significantly different from the estimates of those physicians (14%) who recommended additional noninvasive testing (19 +/- 4%) or those (29%) who recommended medical therapy (28 +/- 26%) (difference not significant). Thus, cardiologists accurately estimated prognosis following AMI, but they were less accurate in assessing high risk than low risk, and their management decisions correlated poorly with their risk assessments.