Registers of the management of infarction can complement information obtained from randomised trials evaluating the methods and practice of treatment. In order to do this, the quality of the registers must be assured, and in particular the accuracy of the recorded cases. The objective of this study was to evaluate the accuracy of a register for the in-hospital and pre-hospital management of acute coronary syndromes with ST segment elevation of less than 12 hours' duration. Using a capture-recapture method, the study compared cases in the register with eligible cases present in the hospital and emergency ambulance service databases at two establishments, giving a recruitment rate of 61%. The rate of accuracy was estimated at 84% (95% CI [82 ; 86]).The independent factors associated with failure of notification were female sex (ORa=6.65 [2.04-21.69]), presentation at nights, weekends or bank holidays (ORa=4.13 [1.33-12.85]), direct admission to hospital without passing by the emergency ambulance service (ORa=2.85 [1.03-7.69]), primary angioplasty (ORa=6.18 [1.60-23.79]) and the absence of reperfusion (ORa=40.38 [6.21-262.40]). With more than 80% accuracy, the results produced by the register are robust. The selection bias linked to the under-representation of certain subgroups, while real, has only a marginal impact on estimates derived from the register. Factors associated with failure of notification should be taken into account when operating such a register.