For some patients with coronary artery disease (CAD), bypass operations prolong life. Angiograms, incurring some risk and considerable expense, are prerequisites to surgical therapy; they delineate the region and extent of disease. However, many people who complain of chest pain do not have disease that can be benefited by operation. Therefore, tests that will safely and economically select the appropriate individuals for angiography are most welcome. Yet, if the preliminary tests falsely declare affected people to be free of CAD, they will deny these patients angiography, and, consequently, surgical treatment that would prolong their lives. Decision analysis determines that a false-negative rate of less than 2% is necessary for tests preliminary to angiography if the average survival of patients is not to be shortened. No currently used procedure has attained this sensitivity. Radionuclide ventriculography approaches this precision, but its sensitivity must be sustained in more broadly based studies.