The study objective was to assess the widespread applicability of ST/HR slope for the modified Bruce exercise test using a computerized electrocardiogram (ECG); compare the usefulness of the ST/HR slope with standard ECG criteria in detection of coronary artery disease (CAD) and identification of three-vessel or left main CAD; and then develop a new, modified ST/HR score (MSHS) for improving the diagnostic accuracy of ST/HR slope. The studies were retrospective and prospective in design, conducted in referral-based cardiology clinics at a national cardiovascular center. A selected sample of 142 patients underwent exercise ECG and coronary angiography, as did a normal control group of 402 patients who were apparently free from CAD. Sixty three other patients who underwent coronary angiography were also studied prospectively. No limitations of medical treatment were exacted for the test except digitalis treatment. Linear regression analysis, from which ST/HR slope was derived, was done with seven measurements of HR and ST displacement at 60 ms from J point in leads a VF and V5 during 6 min before the end of exercise. MSHS was derived from a multiple regression model with peak HR (% maximum HR), ST index (ST depression + ST slope), and ST/HR slope. Although the usual ST criteria (sensitivity = 63 percent, specificity = 73 percent), ST index (71 percent, 80 percent), and ST/HR slope (70 percent, 97 percent) were equally accurate in detection of CAD, MSHS showed significantly improved sensitivity (88 percent) with similar specificity (81 percent). In identification of three-vessel or left main CAD, when compared with other criteria, ST/HR slope and MSHS provided improved diagnostic accuracy: sensitivity (74 and 78 percent, respectively), specificity (88 and 93 percent) and overall test accuracy (85 and 89 percent). The improved accuracy of ST/HR slope and MSHS was prospectively validated in 63 other patients. ST/HR slope was applicable to computerized ECG data for the standard treadmill test, and showed improved accuracy in detection of three-vessel or left main CAD. The new, modified ST/HR score more accurately predicted not only the presence but also the severity of CAD.