The diagnostic contribution often common exercise tolerance test (ETT) measures compared with coronary angiography was studied in 62 symptomatic females (mean age = 53 ± 9 years). Logistic regression revealed that maximal ST-segment depression, the percent of predicted maximal heart rate achieved, and test chest pain all contributed unique predictive information and formed a model generating probabilities for coronary disease (CAD). Using a predicted probability for the presence of CAD of 0.50 as a cutpoint, test accuracy was markedly improved (sensitivity = 73%, specificity = 94%, and overall correct classification rate = 90%) over the standard ST response. We conclude that multivariate analysis using these three easily assessed ETT measures provides superior discrimination between symptomatic women with and without CAD when compared to changes in the ST-segment alone.