Some patients with coronary artery disease (CAD) and exercise-induced myocardial ischemia demonstrate no change or a paradoxical increase in systolic blood pressure (SBP) during recovery following exercise. Previous studies have investigated the significance and clinical usefulness of analysis of recovery SBP response in detecting CAD, but conflicting data have been reported. Different protocols were used for the time of SBP recording and either bicycle or treadmill testing. We studied the exercise response in 64 male patients investigated for CAD who underwent symptom-limited treadmill stress testing during electrocardiographic monitoring and serial recording of blood pressure. Forty-three patients showed on or more stenoses of at least 70% at angiography (CAD). Twenty-one patients with normal coronary tree or slight lesions served as controls. The sensitivity (true positive/all CAD patients), specificity (true negative/all CAD-free patients), and the correct classification rate (correct diagnoses/all subjects) were assessed by standard ST segment analysis and two recovery SBP ratios calculated by dividing the first minute recovery SBP by the immediate postexercise value (RR/R) or by the true peak exercise value (RR/P). ST segment analysis achieved 53% sensitivity, 57% specificity, and 54% correct classification, the RR/R ratio achieved 73%,23%, and 60%, and the RR/P ratio 53%, 71%, and 59%, respectively. There were significant differences in results using these ratios. Time of SBP recording generated discrepancies in recovery SBP ratios. Therefore, differences in the timing of SBP measurement may generate conflicting clinical indications.