The ST-segment/heart rate (ST/HR) slope has markedly improved the accuracy of the exercise electrocardiogram for the identification and quantification of coronary artery disease. However, clinical use of this technique has been limited by time-consuming manual ST-segment measurements and calculator-based linear regression analyses after testing. To assess the accuracy of on-line computerized measurement of ST-segment depression and recently implemented ST/HR slope algorithms, 50 patients exercised with a Marquette Electronics CASE 12 system and 50 patients exercised with a Quinton Q5000 system were evaluated. Computerized ST-segment measurements, to the nearest 10 microV, were compared with averaged manual measurements from the raw signal, to the nearest 25 microV, at peak exercise in leads II and CM5. The CASE 12 and Q5000 algorithms for selection of data and calculation of the maximal ST/HR slope were compared with the ST/HR slope calculated from end-stage data points according to a standard off-line procedure. Manual ST-segment measurements correlated highly with CASE 12 and Q5000 determinations (r = 0.996 to 0.998). The slopes of their regression lines approached unity (range 0.98 to 1.02), and the standard errors of the estimate ranged from 12 to 15 microV. Each on-line program selected data points and determined ST/HR slopes in complete accordance with standard calculations. It is concluded that computer-based measurement of the ST/HR slope can be performed with a high degree of accuracy.