In a large Phase III double-blind randomised clinical trial, the CAPRIE study (Clopidogrel versus Aspirin in Patients at Risk of Ischemic Events), two anti-platelet agents were compared in secondary prevention of ischemic events (myocardial infarction, ischemic stroke, vascular death) in atherothrombotic patients. The study was not designed to collect health economics parameters. OBJECTIVES: we needed to assess the incremental cost-effectiveness- ratio of the new anti-platelet agent in the main european countries and to test various hypotheses. METHODS: Clinical outcomes : A Markov model designed with several clinical states combined the rates of all ischemic events (fatal and non fatal, primary and subsequent) reported in the CAPRIE study. The rates of adverse events were also taken into account. Survival data : the remaining life expectancy of the patients who experienced an event was not directly estimated from the CAPRIE study as the mean follow-up of the patients was only two years. The survival estimates was approached by identifying ,in the Framingham study ,cohorts of patients similar to the CAPRIE patients. Resources use and Costs : decision trees about inpatient and outpatient management of myocardial infarctions and ischemic strokes were built from available data (literature, data base, expert opinion) and were locally validated. RESULTS: The economic model allowed to calculate the incremental cost-effectiveness- ratio of the new anti-platelet agent for the basic scenario (CAPRIE study) and for other scenarios (changes of the time horizon, considering that the comparator could not be the reference treatment for all the patients in a real life situation, high risk patients analysis). CONCLUSION: Because they are protocol driven, the Phase III double-blind randomised clinical trials do usually not reflect what would happen in the real life. Modeling, which allows to perform simulations, could be a way of correcting.