The temporal distribution of the diagnostic information for the detection of coronary artery disease (CAD) provided by exercise-induced electrocardiographic (ECG) ST-segment amplitude changes in different ECG leads in men and women has not been fully investigated. To shed further light in this area, 1877 electrocardiograms selected from 8322 patients undergoing a routine exercise test on a bicycle ergometer were evaluated. ST-segment amplitude and the difference between heart rate-matched recovery and exercise ST-segment amplitudes (ST/HR difference) were measured. Coronary artery disease was verified angiographically in 669 patients and excluded in 1208 patients by angiography (n = 119), by myocardial scintigraphy (n = 250), or on clinical grounds (n = 839). The diagnostic performance of the 2 ECG methods used was assessed by constructing receiver operating characteristic curves for each sampling point every 12 seconds during 10 minutes of recovery as well as the last 4 minutes of exercise for the ST-segment amplitude. ST-segment amplitude performed better after exercise than during exercise and best within the first 2 minutes of recovery. Its diagnostic ability did not differ from the ST-amplitude hysteresis assessed by the difference between recovery ST-segment amplitude and exercise ST-segment amplitude at matched heart rate. Both methods performed better in men and the diagnostic information appeared mainly in leads I, -aVR, II, V 4 , V 5 , and V 6 . The best discrimination of CAD is provided by analysis of ST-segment amplitude changes in 6 specific leads early during the recovery phase. This information should be targeted by exercise ECG diagnostic methods.