In the GUSTO-I ECG ischaemia monitoring substudy, 1067 patients underwent continuous ST segment monitoring, using vector-derived 12-lead (406 patients), 12-lead (373 patients) and 3-lead Holter (288 patients) ECG recording systems. Simultaneous angiograms at 90 or 180 min following thrombolytic therapy were performed as a part of the prospective study in 302 patients. Infarct vessel patency was established as TIMI perfusion grades 2 or 3 and occlusion as TIMI perfusion grades 0 or 1. Coronary artery patency was predicted from ST trends up to the time of angiography. Predictive values at 90 and 180 min after the start of thrombolysis were 70% and 82% for patency and 58% and 64% for occlusion, respectively. In retrospect, accuracy appeared greatest (79-100%) in patients with extensive ST segment elevation (> or = 400 microV), if both speed of ST recovery and extent of ST segment elevation were taken into account. Although the three recording systems differed considerably in signal processing, no significant difference in accuracy was demonstrated among these systems. We conclude that continuous ECG monitoring may help select high risk patients without apparent reperfusion who may benefit from additional reperfusion therapy. As ST recovery may occur early after the start of thrombolytics and accuracy of the test is related to peak ST levels, the use of on-line ECG monitoring devices on emergency wards and cardiac care units is recommended.