Abstract Objective The study objective was to determine how best to use high-sensitivity cardiac troponin T (hscTnT) to diagnose myocardial infarction. Methods A total of 358 patients presenting with acute coronary syndromes sampled at admission and 2, 4, and 6 to 8 hours. Both contemporary cardiac troponin T (cTnT) and hscTnT were measured. Patients were classified with conventional cTnT values by independent investigators. Myocardial infarction required a cTnT value ≥99th reference percentile and a ≥20% change. Results Seventy-nine patients had non-ST-segment elevation myocardial infarction, 105 patients had unstable angina, and 174 patients had nonacute coronary syndromes. A cTnT cutoff at the 10% coefficient of variation value missed 14.5% of infarctions. hscTnT had a sensitivity at admission of 89.9%, but specificity was only 75.1% because of elevations in 45.3% and 25.3% of those with unstable angina and nonacute coronary syndromes, respectively. The optimal value for myocardial infarction diagnosis with hscTnT was 25 ng/L at admission and 30 ng/L during serial sampling. All infarctions were diagnosed within 4 hours, with a time saving of 11 and 68 minutes compared with a cTnT value at the 99th reference percentile value and a cTnT value at a coefficient of variation of 10%. By using the 99th percentile of hsTnT plus a ≥20% change, 25 additional infarctions were identified. With these included, the optimal cutoff decreased to 12 ng/L at admission and 13 ng/L over time, but time to diagnosis increased. Conclusions The gold standard used to diagnose myocardial infarction makes a major difference in the results. When myocardial infarction is diagnosed using hscTnT 99th percentile values with a 20% change, more are identified, diagnosis is delayed, and the optimal value for use is reduced.