The interrelations of clinical, exercise test, and angiographic variables and their relative values in predicting specific clinical outcomes after myocardial infarction have not been fully established. Of 302 consecutive stable survivors of infarction, 262 performed a predischarge submaximal exercise test. In the first year after infarction patients with a "positive" exercise test were 13 times more likely to die, 2.8 times more likely to have an ischaemic event, and 2.3 times more likely to develop left ventricular failure than patients with negative tests. Patients with positive exercise tests underwent cardiac catheterization. Features of the history, 12 lead electrocardiogram, in-hospital clinical course, exercise test, and left ventricular and coronary angiograms that predicted these clinical end points were identified by univariate analysis. Then multivariable analysis was used to assess the relative powers of all variables in predicting end points. Certain features of the exercise test remained independent predictors of future ischaemic events and the development of overt left ventricular failure, but clinical and angiographic variables were more powerful predictors of mortality. Because the exercise test is also used to select patients for angiography, however, the results of this study strongly support the use of early submaximal exercise testing after infarction.