Available noninvasive techniques for identifying patients with failed epicardial reperfusion after fibrinolytic therapy are limited by poor accuracy. It is unknown whether combining multiple noninvasive predictors would improve diagnostic accuracy and facilitate identification of candidates for rescue percutaneous coronary intervention. In the Thrombolysis In Myocardial Infarction (TIMI) 14 trial, we evaluated the ability of ST-segment resolution (n = 606), chest pain resolution (n = 859), and the ratio of 60-minute/baseline serum myoglobin (n = 308) to identify patients with angiographic evidence of failed reperfusion 90 minutes after fibrinolysis. Three criteria were prospectively defined: <50% ST resolution at 90 minutes, presence of chest pain at the time of angiography, and myoglobin ratio <4. Patients who met any individual criterion were more likely to have less than TIMI 3 flow and an occluded infarct-related artery (TIMI 0/1 flow) than those who did not meet the criterion (p <0.005 for each). When the 3 criteria were used together (n = 169), patients who satisfied 0 (n = 29), 1 (n = 68), 2 (n = 51), or 3 (n = 21) of the criteria had a 17%, 24%, 35%, and 76% probability of failing to achieve TIMI 3 flow (p <0.0001 for trend), a 0%, 6%, 18%, and 57% probability of an occluded infarct-related artery (p <0.0001 for trend), and a 0%, 1.5%, 2.0%, and 9.5% rate of 30-day mortality (p = 0.05 for trend), respectively. Use of the criteria in combination increased positive predictive values without decreasing negative predictive values. In conclusion, ST-segment resolution, chest pain resolution, and early washout of serum myoglobin can be used in combination to aid in the early noninvasive identification of candidates for rescue percutaneous coronary intervention.