The ability of radionuclide variables obtained at rest and at peak exercise to discriminate the number of stenosed (greater than or equal to 70% luminal diameter narrowing) major coronary arteries was evaluated in 296 patients undergoing supine exercise radionuclide ventriculography. Stepwise linear discriminant analysis of the data from the first 200 patients identified a significant (p less than 0.001) discriminatory combination. Application of this function to the remaining 96 patients provided correct classification of arteriographically determined zero, one, two, and three stenosed arteries in 59%, 18%, 14%, and 60% of cases, respectively. The discriminant function classified minimal stenoses (zero or one artery) and multivessel stenoses (two or three arteries) correctly by arteriography in two thirds of cases in each group. Arteriographic presence of three stenoses was unlikely in those classified as having no stenosis, and absence of stenosis was rare in those classified as having three stenoses. Exercise radionuclide ventriculography is most helpful in identifying minimal and multivessel coronary disease rather than number of stenosed major coronary arteries.