To evaluate the clinical utility of perfusion scintigraphy in patients with markedly pesitive exercise ECGs, we studied 94 consecutive patients with markedly positive exercise ECGs; 74 of them were also studied with scintigraphy. Patients undergoing scintigraphy had an intermediate pretest likelihood of coronary disease and were divided into two groups: those with reversible defects involving one complete area or aspects of multiple vascular areas (group 1, 38 patients), and those with normal scintigrams and reversible defects in a limited aspect of one vascular area, isolated fixed defects, or both (group 2, 36 patients). Among all demographic and exercise variables, only a hypotensive or blunted blood pressure response and scintigraphic lung uptake or cavitary dilation, although insesitive, were more frequent in group 1 (all p < 0.05). All 36 patients in group 1 and 14 of 18 in group 2 who underwent coronary angiography had significant coronary lesions; 31 in group 1 but only seven in group 2 had multivessel disease (p < 0.05). Subsequently 32 patients in group 1 had revascularization compared with only two patients in group 2. Only one cardiac event was noted among 34 patients in group 2 who were followed for a mean of 38 months while they were being treated medically. However, four of nine patients in group 1 initially treated medically required late revascularization because of clinical progression of disease, and one patient died (p < 0.05). Compared with patients having scintigraphy, patients not imaged had a higher pretest likelihood of coronary disease, a higher incidence of angina, unstable angina, and induced angina, with a lower exercise time and time to ST depression (p < 0.05). All underwent angiography, and 16 had multivessel disease. Not all patients with markedly positive exercise ECGs were at similarly high coronary risk. Some with high-risk coronary anatomy were identified without the use of scintigraphy. In others, where diagnosis and prognosis were less clear, scintigraphy aided in the diagnosis and accurately identified a low-risk subgroup as did no other parameter.