This study compared the accuracy of rest and exercise gated equilibrium technetium ventriculography with exercise thallium imaging in 50 consecutive male patients undergoing routine coronary angiography for the evaluation of chest pain. No patients were excluded on the basis of prior myocardial infarction, nature of angiographically defined coronary disease or symptoms. Antianginal therapy was continued in all patients. Eight patients had normal coronary arteries, 9 had single vessel, disease, 20 had double vessel disease and 13 had triple vessel disease. Sixteen patients had previously documented myocardial infarction. Using exercise radionuclide ventriculography, 34 patients with coronary disease were detected resulting in a sensitivity of 81%; 6 patients with normal coronary arteries had normal scans, a specificity of 75%, with a predictive accuracy of 80%. In comparison, thallium imaging detected 42 patients with coronary disease resulting in a sensitivity of 100%. Six patients with normal coronary arteries had normal thallium images resulting in a specificity of 75% and a predictive accuracy of 96%. These results suggest that exercise thallium imaging is a more accurate investigation than exercise equilibrium radio-nuclide ventriculography and is the investigation of choice in the noninvasive detection of coronary artery disease.