Women have a notoriously high rate of false positive exercise test results. Since the exercise ST segment response has low specificity in predicting CAD in women, we examined additional exercise parameters in 200 women with a history of chest pain compatible with angina and having ST segment depression greater than or equal to 1 mm recorded during a Bruce treadmill test. All subsequently had coronary arteriography. Two groups were compared: group A (n = 80) with CAD (greater than or equal to 70 percent stenosis of one or more coronary artery) and group B (n = 120) with angiographically confirmed normal coronary arteries (normal or minimal placquing). The exercise criteria analyzed included: (1) chest pain during exercise, (2) percent target heart rate, (3) extent of ST shift, (4) morphology of the ST segment slope, (5) time to normalization of the ST segment, and (6) total exercise duration. Multivariate analysis (using a stepwise logistic regression model) identified four independent exercise variables associated with the likelihood of CAD: (absence of MVP, p = .003; exercise duration less than 5 min, p = .02; ability to reach target heart rate, p = .027; time to ST normalization greater than or equal to 6 min, p less than .001). False positive exercise test results were more likely to occur when the following exercise test variables were present: ability to exercise to stage 3 of the Bruce protocol and a rapid (less than or equal to 4 minutes) normalization of ST shift after cessation of exercise. Attention to these additional exercise variables allows more careful selection of women requiring more definitive (and expensive) testing.