The results of stress testing in patients being assessed for coronary artery disease have led to conflicting claims and conclusions. It seems reasonably clear that stress testing is of considerable predictive value inepidemiologic studies;i.e., patients manifesting or developing a positive exercise test have a much higher probability of subsequently experiencing coronary events (angina pectoris, myocardial infarction, or coronary death) than those individuals with normal stress tests. Moreover, the risk seems to be related to the degree of S-T segment depression. In contrast, despite earlier claims regarding thediagnosticaccuracy with which single-load stress tests could predict the presence or absence of ischemic heart disease in individual patients, studies correlating the results of testing with the degree and extent of angiographically demonstrated coronary artery disease have not consistently shown either acceptable sensitivity or specificity. Although on theoretic grounds it would be anticipated that multistage stress tests would be superior to single-stage protocols, this has not been borne out in the few published studies in which the results are correlated with angiography. Thus, the available evidence suggests that there remain appreciable numbers of patients with documented coronary artery disease in whom no ECG abnormalities are detected despite relatively intense levels of exercise, and an appreciable number of patients with abnormal ECG responses but normal coronary arteries. Stresses other than exercise have been and are being utilized in the evaluation of patients with chest pain; however, no single test as yet offers the desired specificity and sensitivity.We conclude that, in patients with typical angina pectoris or with chest pain which clearly does not resemble angina pectoris, stress testing appears to be superfluous since it provides little additional information beyond that which may be obtained from the patient's history. It is in those patients presenting with atypical anginal syndromes that a reliable noninvasive test would be of great value. Currently available technics are neither sufficiently sensitive nor specific to satisfactorily aid in solving this problem.