ST-segment resolution is used to classify the response to reperfusion therapy in acute myocardial infarction, but the possibility to predict outcome in individual patients is unclear, particularly in the setting of primary percutaneous coronary intervention (PCI) and abciximab therapy. We studied 213 patients who underwent successful revascularization with PCI. Maximal ST-segment elevation was measured before and 30 minutes after PCI. Patient outcome was defined on the basis of infarct size and left ventricular ejection fraction (EF) as derived from gated single-photon emission computed tomography that was acquired 1 month after infarction. Patients who had > or =50% ST resolution showed a smaller infarct (15.1 +/- 13.6% vs 19.9 +/- 15.7%, p < 0.05) but not a higher left ventricular EF (48.7 +/- 12.3% vs 45.2 +/- 11.8%) than did patients who had <50% resolution. According to cluster analysis of infarct size and left ventricular EF, 132 patients had favorable outcome (central values: infarct size 7.5%, left ventricular EF 55%) and 81 did not (central values: infarct size 30%, left ventricular EF 36%). Using receiver-operating characteristic curve analysis, the optimal ST-resolution cutoff was >60%, with 77% sensitivity and 51% specificity for predicting favorable outcome. ST-segment elevation < or =4.5 mV before PCI was 80% sensitive and 48% specific, and ST-segment elevation < or =1 mV after PCI was 74% sensitive and 60% specific for predicting favorable outcome. In conclusion, in the setting of primary PCI and abciximab therapy, ST-segment elevation resolution requires a high threshold (>60%) to effectively classify patients; the capability of ST-segment analysis to predict patient outcome is limited, with ST-segment elevation after PCI showing the best compromise between sensitivity and specificity.