When the measurement of a long-term outcome is necessary, selecting evidence-based time horizons according to pharmacoepidemiology data is crucial. OBJECTIVE: To illustrate the effect of assumed time horizon for the interventions in pharmacoeconomic models on measured outcomes. METHODS: The benefit of reducing LDL-C was incorporated into a model to calculate reduction in cardiovascular events and resulted economic outcomes. Data for LDL-C reduction from a head-to-head RCT [Am Heart J 2002;144:1044-51]; rosuvastatin (starting 5 mg) versus atorvastatin (starting 10mg) with up-titration doses were incorporated into the model; and distribution of cardiovascular risk for users [N = 100,000, duration five years] in Canadian population [Clin Invest Med 2007;30:E63-E69] were assumed. To find out the effect of time horizon on economic evaluation of therapeutics, the component of five years was changed to ten years time horizon. RESULTS: Using five years duration of therapy, rosuvastatin and atorvastatin can prevent 9505 and 8702 cardiovascular events (non-fatal MI and stroke), respectively. Reduction in non-fatal MI and stroke can be translated to $252,300,392 (CDN) and $230,980,624 direct cost savings, respectively ($288,871,921 and $158,510,416 total net-benefit). With ten years assumption for statin therapy, rosuvastatin and atorvastatin can prevent 25948 and 22190 cardiovascular events, respectively. The prevention of cardiovascular events according to the model based on ten years time horizon were calculated 2.73 and 2.55 times higher than the five years based model for rosuvastatin and atorvastatin, respectively. CONCLUSION: This simulation study illustrates the effect of incorporated time horizons in pharmacoeconomic models on the resulted outcomes. Therefore, considering an evidence-based time horizon for the model is essential. For example, in this study Canadian community-based clinical practice data reported a median of approximately five years of statin therapy for the patients. Therefore, a time horizon of five years was assumed to be an evidence-based time horizon for the model.