Although quantification of exercise thallium images has been previously reported, the relative value of different imaging variables for detection of coronary artery disease has not been analyzed in a large group of patients with cardiac catheterization data. Regional initial thallium uptake, redistribution and clearance on thallium study were measured in 325 patients also undergoing cardiac catheterization (281 patients with and 44 patients without coronary artery disease). Normal values were defined in 55 other clinically normal subjects. When five myocardial segments were analyzed in each view, the respective values for sensitivity and specificity were 95 and 50% for initial thallium uptake, 60 and 87% for redistribution and 74 and 66% for clearance. Initial thallium uptake was the most sensitive but least specific (p less than 0.001), whereas redistribution was the least sensitive and most specific (p less than 0.001). Using stepwise logistic regression analysis, the best correlate of coronary artery disease was initial thallium uptake. Addition of redistribution to a mathematical model of the probability of coronary artery disease did not alter sensitivity, but increased specificity from 50 to 70% (p less than 0.001). Once initial uptake and redistribution were considered, myocardial thallium clearance provided no additional improvement in the correlation. Excluding the two basal segments in each view from the analysis increased the specificity from 70 to 80% (p less than 0.001) without affecting sensitivity. Of the 15 patients (5%) with coronary disease not detected using this approach, none had left main disease and 10 (67%) had one vessel disease. A combination of variables derived from quantification of exercise thallium images provides a superior sensitivity and specificity for the detection of coronary artery disease compared with the use of a single variable.