Although the time course of ST segment depression after exercise has been related to the presence and severity of coronary artery disease, recovery-phase patterns of ST segment depression with reference to changing heart rate have not been quantified. We have found distinct recovery loop patterns of ST segment depression that distinguish subjects without coronary disease from patients with coronary artery disease when ST segment depression is examined in the heart rate domain. Continuous plots of ST segment depression and heart rate were constructed throughout treadmill exercise and recovery in 100 clinically normal subjects, in 124 patients with coronary artery disease proven by catheterization, and in 17 patients with no significant coronary disease at catheterization. Among clinically normal subjects, 95% (95 of 100) had normal (clockwise) rate-recovery loops, and 5% (five of 100) had abnormal (counterclockwise) rate-recovery loops. In these normal subjects, the resulting 95% specificity of a normal rate-recovery loop was similar to the 93% (93 of 100) specificity of standard end-exercise ST segment depression criteria. Among patients with coronary disease proven by angiography, 93% (115 of 124) had abnormal (counterclockwise) rate-recovery loops, and 7% (nine of 124) had normal rate-recovery loops. In contrast was the significantly lower 74% (92 of 124) sensitivity of standard ST segment criteria (p<0.001 vs. the rate-recovery loop).Specificity of a normal rate-recovery loop (71%, 12 of 17) and standard ST segment depression criteria (71%, 12 of 17) were similar in the patients with normal coronary arteries at angiography. We conclude that the pattern of ST segment depression as a function of heart rate during exercise and recovery can markedly enhance the accuracy of the exercise electrocardiogram for the identification of coronary artery disease. (Circulation 1989;80:533-