BACKGROUND Before any new diagnostic test is accepted in clinical practice, such a test should be compared with established diagnostic tools in an appropriately large series of patients encompassing the complete spectrum of challenges to which the test is exposed. The aim of the present study was to assess the relative diagnostic and prognostic accuracies of high-dose dipyridamole echocardiography (two-dimensional echocardiographic monitoring during dipyridamole infusion up to 0.84 mg/kg over 10 hours) versus maximal symptom-limited bicycle exercise ECG test in patients with angina. RESULTS We studied 429 consecutive in-hospital patients who met the following inclusion criteria: history of chest pain, off antianginal therapy for at least 2 days (1 week for beta-blockers), no previous myocardial infarction and/or obvious regional left ventricular dyssynergy of contraction (akinesis or dyskinesis) at baseline, and acceptable acoustic window under resting conditions. All patients underwent dipyridamole echocardiography and exercise ECG--on different days and in random order--within 1 week of coronary angiography (which was performed independent of test results) and were followed up for 37.8 +/- 14 months (range, 1 to 73 months). Criteria of positivity were for dipyridamole echocardiography, a transient regional dyssynergy absent in the baseline examination; for exercise ECG, an ST-segment shift of > or = 0.1 mV from baseline; and for coronary angiography, a luminal reduction of > or = 75% in at least one major coronary vessel (50% for left main). There were 183 patients without and 246 with coronary artery disease; 132 had one-, 70 had two-, and 44 had three- and/or left main vessel disease. The specificity was higher for dipyridamole echocardiography than for exercise ECG (90% versus 51%, P < .001). The overall sensitivity of dipyridamole echocardiography was similar to that of exercise ECG (75% versus 74%, P = NS), with no significant differences in the subset with one- (67% versus 69%, P = NS), two- (79% versus 77%, P = NS), or three- (93% versus 86%, P = NS) vessel disease. During the follow-up, there were 20 deaths, 13 nonfatal myocardial infarctions, and 126 revascularization procedures. In the univariate analysis, dipyridamole resulted in higher chi 2 values than did exercise stress testing. A Cox forward stepwise survival analysis identified the dipyridamole time as the most powerful prognostic predictor of death (chi 2 = 19.4, P < .0001) of all invasive and noninvasive parameters. The dipyridamole time also provided independent and additional prognostic information when it was adjusted for age, diabetes, resting ECG, and exercise stress test according to a modified, interactive stepwise procedure. This is true when death only, death and myocardial infarction, and death, myocardial infarction, and revascularization procedures were considered end points. CONCLUSIONS In patients with no previous myocardial infarction and good resting left ventricular function, compared with exercise ECG, dipyridamole echocardiography has a similar sensitivity and a higher specificity for the noninvasive detection of angiographically assessed coronary artery disease. Dipyridamole echocardiography also provides information in addition to that provided by exercise ECG for predicting death, infarction, and all events when the presence as well as the timing, severity, and extension of dipyridamole-induced wall motion abnormalities are considered.