In a substantial proportion of patients with suspected myocardial infarction, biochemical markers are needed for clinical decision-making at the time of admission, because electrocardiographic (ECG) recordings are inconclusive. We have assessed the usefulness for exclusion of myocardial infarction at admission of the newer markers creatine kinase MB (CK-MB) mass concentration, troponin T, and myoglobin in comparison with the routinely used markers creatine kinase (CK) and CK-MB activity. 290 consecutive patients were enrolled. Acute myocardial infarction was diagnosed on the basis of clinical history, ECG criteria, and time-dependent changes in CK and CK-MB activity. 153 patients had definite acute myocardial infarction. Troponin T had the highest sensitivity for prediction of acute myocardial infarction; high concentrations (above the upper reference limits) were found in 98 (64%) of the patients with infarctions compared with 92 (60%) for CK-MB mass concentration, 76 (50%) for myoglobin, 61 (40%) for CK activity, and 53 (35%) for CK-MB activity. However, troponin T also had the highest "false-positive" rate; of 137 patients without myocardial infarction, 36 (26%) had high troponin T concentrations. Sensitivity, specificity, and positive and negative predictive values were calculated in relation to time between onset of chest pain and hospital admission. Although CK-MB mass concentration was, by a small margin, the best marker in patients admitted within 8-10 h of onset of chest pain, all the markers had negative predictive values too low to allow exclusion of acute myocardial infarction at admission in patients with symptoms suggestive of myocardial infarction of less than 10 h duration.