The effect of nonstandardized creatine kinase (CK)-MB assays on the assessment of myocardial infarction (MI) end points in multicenter international trials has not been evaluated. We compared the site-reported and corresponding core laboratory CK-MB measures from 5 countries participating in the Superior Yield of the New Strategy of Enoxaparin, Revascularization, and Glycoprotein IIb/IIIa Inhibitors (SYNERGY) trial. Samples for CK-MB were collected locally, with corresponding samples sent to a core laboratory at enrollment and after recurrent ischemic events, percutaneous coronary intervention, or coronary artery bypass grafting. The measured values were compared to the reported assay upper limits of normal (ULN) used at the site (or core laboratory for the core laboratory samples). The CK-MB results were available locally and from the core laboratory for 913 patients, constituting 4,693 time-matched laboratory values. The agreement between the core and site laboratory CK-MB/ULN ratio was moderate (concordance correlation coefficient 0.45) and varied considerably by geographic location and site. The CK-MB values were elevated (>or=2 times the ULN) by the core laboratory but normal (<2 times the ULN) by local standards in 708 instances (15%). There were 162 MI end points according to the core laboratory values versus 91 MI end points using the site-reported CK-MB data (kappa statistic 0.48). Compared with patients with no MI by the core or site laboratory values, patients with MI, as determined by both the core and the site laboratories, had significantly lower unadjusted 1-year survival rates (80.6% vs 93.5%, p <0.0001). Patients with MI, as determined by the core laboratory but not by the site laboratory, showed a trend toward a lower 1-year survival rate (89.8% vs 93.5%, p = 0.20). In conclusion, a substantial variation in CK-MB ratios and MI outcomes between the site and core laboratory data was observed in the SYNERGY trial. More MI outcomes were identified by the core laboratory, and patients with MI as defined by core laboratory data had lower 1-year survival, making these events potentially clinically important.