Exercise electrocardiograms of 20 patients were analyzed using a customized software exercise electrocardiographic program and compared to measurements made by two cardiologists performing independent interpretations. The computerized program requires identification of the PQ junction, J point, and tracing of the ST-segment in three consecutive beats. The proportion of variance for J point, and ST 80 measurements was 0.93 and 0.90, respectively, when the same electrocardiogram was processed twice and analyzed by two separate cardiologists. The same 20 exercise electrocardiograms were analyzed by two other experienced cardiologists without computerized measurements. The proportion of variance was less at 0.73 and 0.76 for the J point and ST 80 measurements, respectively. The average amount of time required for the cardiologist to over read the computerized measurements was 2.7 +/- 1.5 minutes per ECG as compared to 20.7 +/- 11 minutes for the cardiologists who did not have computer-assisted measurements (p less than 0.0001). Thus, off-line computerized exercise electrocardiographic interpretation is highly reproducible, accurate, time-sparing for cardiologist over read function, and suitable for use in large-scale clinical trials.