Percutaneous coronary intervention (PCI) is considered the best treatment for acute myocardial infarction with ST segment elevation (STEMI), but it is difficult to deliver.To report on long-term mortality predictors in a registry based on a 'hub and spoke' model, according to the initial strategy: thrombolysis followed or not by PCI, invasive strategy followed or not by primary PCI and no reperfusion.From May 2001 to June 2003, 514 patients (mean age 67 +/- 12) with STEMI onset less than 12 h ( or =6 leads, Killip class >1 and previous STEMI). Mean TRS score was 4.0 and 53% of patients met LHRC. Thrombolysis was undertaken in 49% of patients, invasive strategy in 29% and no reperfusion in 22%. The latter had higher TRS (4.9) but only 40% met LHRC. Reperfusion time was significantly longer in patients who underwent PCI as compared with those who underwent thrombolysis (223 vs. 120 min, P < 0.0001). Patients in the thrombolysis group had better risk profiles and underwent emergency or elective revascularization within 30 days in 66% of cases. Overall, long-term mortality rate (36 months) was 23.3%. Both TRS and LHRC identified patients with higher mortality (43 and 32%, respectively). Multivariate analysis showed age, left ventricular ejection fraction and Killip class more than 1 to be significant predictors of mortality (P < 0.0001/P < 0.0001/P = 0.0103), whereas reperfusion strategy and time to treatment were not.An initial strategy of thrombolysis followed by emergency or elective PCI as appropriate is still an option in a setting in which limited resources are available. Decision-making based on risk scores and time from symptom onset lead to proper patient selection and even to foregoing reperfusion without affecting mortality.