assays have been developed which can detect much lower levels of troponin. Methods: 441 patients with chest pain suggestive of acutemyocardial infarction (AMI)were recruited between November 2006 and May 2007; all were followed for 2 years. Serial blood samples were analysed to compare Roche high sensitivity troponin T (hsTNT)with latest generation Abbott troponin I (TNI) and Roche TNT for the prediction of MACE, defined as composite of cardiovascular death, myocardial infarction or revascularization. Results: 91 patients (20.6%) experienced MACE at 2 years. ROC curve AUC for baseline hsTNT was 0.71 (0.66–0.75), TNI was 0.66 (0.61–0.71) and TNT was 0.62 (0.56–0.67). hsTNT out-performs TNT (p= 0.001) but there is no statistical difference between hsTNT and TNI or TNI andTNT.Hazard ratios (HR) forMACEwere 11.4 (2.8–46.9) for hsTNT, 2.9 (1.9–4.6) for TNI and 2.5 (1.6–3.8) for TNT. WhenHRareadjusted foranypatient characteristic shown to be independent predictors ofMACE (in this study independent predictors were age, prior history of ischaemic heart disease and index admission diagnosis of AMI), only hsTNTconveyedadditional risk,HR, 5.2 (1.2–23.1,p= 0.03). Conclusions: hsTNT outperforms conventional assays in prediction of MACE as seen by ROC curve analysis and Cox proportional hazards modelling and is the only troponin to add prognostic utility when adjusted for all baseline characteristics.