We compared cardiac troponins T (cTnT) and I (cTnI) collected within 3.5 h of ischemic symptoms for predicting clinical outcomes in 770 patients. cTnT (cutoff > 0.1 microgram/L) and cTnI (cutoff > 1.5 micrograms/L) were concordant (both positive or negative) in 90.4% of patients. Among discordant results, 66 were cTnT positive and cTnI negative vs 8 who showed the reverse (P < 0.001). Five cTnT-positive and cTnI-negative patients died within 30 days; none who were cTnT negative and cTnI positive died. cTnT showed a slightly greater association (chi 2 = 18.0, P < 0.001) with 30-day mortality than cTnI (chi 2 = 12.5, P = 0.002). The area of the ROC curve for predicting 30-day mortality was significantly larger (Z = 2.08; P = 0.0375) for cTnT, at 0.68 [95% confidence interval (CI) 0.60-0.75], compared with cTnI, at 0.64 (95% CI 0.56-0.72). When cTnI and the electrocardiogram (ECG) were put in a logistic multiple regression model, cTnT added significant information (chi 2 = 8.03, P = 0.045); however, cTnI did not add to a model containing cTnT and the ECG (chi 2 = 0.84, P = 0.657). cTnT provided more information than cTnI for predicting 30-day mortality early after presentation with acute coronary syndromes.