Qualitative interpretation of tomographic and planar scintigrams, a five point rating scale and receiver operating characteristic analysis were utilized to compare single photon emission computed tomography and conventional planar imaging of myocardial thallium-201 uptake in the accuracy of the diagnosis of coronary artery disease and individual vessel involvement. One hundred twelve patients undergoing cardiac catheterization and 23 normal volunteers performed symptom-limited treadmill exercise, followed by stress and redistribution imaging by both tomographic and planar techniques, with the order determined randomly. Paired receiver operating characteristic curves revealed that single photon emission computed tomography was more accurate than planar imaging over the entire range of decision thresholds for the overall detection and exclusion of coronary artery disease and involvement of the left anterior descending and left circumflex coronary arteries. Tomography offered relatively greater advantages in male patients and in patients with milder forms of coronary artery disease, who had no prior myocardial infarction, only single vessel involvement or no lesion greater than or equal to 50 to 69%. Tomography did not appear to provide improved diagnosis in women or in detection of disease in the right coronary artery. Although overall detection of coronary artery disease was not improved in patients with prior myocardial infarction, tomography provided improved identification of normal and abnormal vascular regions, particularly of the left anterior descending and circumflex artery regions. These results indicate that single photon emission computed tomography provides improved diagnostic performance compared with planar imaging in many clinical subgroups, and suggest that it represents the diagnostic imaging procedure of choice in exercise thallium-201 perfusion studies.