The ability of exercise thallium-201 tomographic imaging to predict the presence of left main or 3-vessel coronary artery disease (CAD) was examined in 688 patients who underwent both exercise thallium-201 testing and coronary angiography. Significant differences existed for multiple variables between patients with (n = 196) and without (n = 492) severe left main or 3-vessel CAD. Logistic regression analysis identified 4 variables as independently predictive of left main or 3-vessel CAD. These variables were the magnitude of ST-segment depression with exercise, the number of visually abnormal short-axis thallium-201 segments, the presence or absence of diabetes mellitus, and the change in systolic blood pressure with exercise. Using these variables, patients were classified by nomograms into low-, intermediate- and high-probability groups. Patients at high probability (n = 205) had a 52% prevalence of 3-vessel or left main CAD, whereas those at low probability (n = 170) had only a 12% prevalence. Only 53 patients (29%) with 3-vessel or left main CAD had perfusion abnormalities in all 3 coronary territories. Clinical and exercise parameters provide important independent information in the identification of left main or 3-vessel CAD by exercise thallium-201 tomographic imaging, because thallium scintigraphy alone is suggestive of extensive CAD in few patients.