OBJECTIVES: This study was designed to compare the prognostic value of an abnormal troponin level derived from studies of patients with non-ST elevation acute coronary syndromes (ACS). BACKGROUND: Risk stratification for patients with suspected ACS is important for determining need for hospitalization and intensity of treatment. METHODS: We identified clinical trials and cohort studies of consecutive patients with suspected ACS without ST-elevation from 1966 through 1999. We excluded studies limited to patients with acute myocardial infarction and studies not reporting mortality or troponin results. RESULTS: Seven clinical trials and 19 cohort studies reported data for 5,360 patients with a troponin T test and 6,603 with a troponin I test. Patients with positive troponin (I or T) had significantly higher mortality than those with a negative test (5.2% vs. 1.6%, odds ratio [OR] 3.1). Cohort studies demonstrated a greater difference in mortality between patients with a positive versus negative troponin I (8.4% vs. 0.7%, OR 8.5) than clinical trials (4.8% if positive, 2.1% if negative, OR 2.6, p = 0.01). Prognostic value of a positive troponin T was also slightly greater for cohort studies (11.6% mortality if positive, 1.7% if negative, OR 5.1) than for clinical trials (3.8% if positive, 1.3% if negative, OR 3.0, p = 0.2) CONCLUSIONS: In patients with non-ST elevation ACS, the short-term odds of death are increased three- to eightfold for patients with an abnormal troponin test. Data from clinical trials suggest a lower prognostic value for troponin than do data from cohort studies.