Because patients who fail to achieve reperfusion after thrombolytic therapy remain at high risk for morbidity and mortality, noninvasive measures of infarct-related artery (IRA) patency are needed to identify candidates for rescue interventions. We prospectively studied 444 patients from the Thrombolysis In Myocardial Infarction (TIMI) 14 trial with interpretable baseline and 90 minute 12-lead electrocardiograms. The percent resolution of ST-segment deviation from baseline to 90 minutes was compared with 90-minute IRA TIMI flow grade, as determined in an angiographic core laboratory. Patients with complete (> or = 70%) ST resolution (n = 208; 47%) had a patency (TIMI 2 or 3 flow) rate of 94%, a TIMI 3 flow rate of 79%, and a 30-day mortality rate of 1.0%. Patients with partial (30% to 70%) or no (< or = 30%) ST resolution had significantly lower rates of patency (72% and 68%; p < 0.0001 vs complete ST resolution) and TIMI 3 flow (50% and 44%; p < 0.0001 vs complete ST resolution), and higher 30-day mortality (4.2% and 5.9%; p = 0.01 vs complete ST resolution). With use of electrocardiographic criteria alone, approximately 50% of patients can be classified as having a high (94%) probability of IRA patency and a very low risk for mortality. Angiography to determine patency of the IRA may be unnecessary in these patients. In patients without complete (> or = 70%) ST resolution, the IRA is still likely to be patent, and additional information from clinical variables or serum markers may help to identify candidates for coronary angiography. Patients with persistent ST elevation despite a patent IRA are at increased risk for mortality, likely due to extensive microvascular and tissue injury.