OBJECTIVE: To compare incidences of cardiovascular disease (CVD) in general and myocardial infarction (MI) specifically between new users of different statins in daily practice. DESIGN AND METHODOLOGY: Retrospective observational cohort study. Datawere obtained from the PHARMO Record Linkage System; the PHARMO database contains pharmacy dispensing records of 3 million patients in the Netherlands registered with community pharmacies, linked to hospitalisation records. The participants were new statin users in the period 2000-2005, excluding patients hospitalised for CVD events in the year prior to start of statin use. MAIN OUTCOME MEASURES: Adjusted hazard ratios of hospitalisations for CVD (including any type of ischemic heart disease, stroke, orrevascularisation procedure) in general, or MI in particular, occurring during use of the initial statin within two years of start of therapy, comparing users of different statins. RESULTS: The mean follow-up duration of 76,147 new statin users (14,530pravastatin, 27,752 simvastatin, 25,777 atorvastatin, 8088 rosuvastatin) was 55 weeks. Incidence rates of CVD and MI per 100 person years ranged from 0.75 and 0.15 for rosuvastatin to 1.72 and 0.29 for pravastatin. Rosuvastatin users had a lower incidence rate of CVD compared to other statin users in general (28% lower), and simvastatin (29% lower) and pravastatin users (40% lower) in particular. The difference with atorvastatin was not statistically significant. Since this was not a prospective randomised study, there is the potential for unobserved risk factors to be responsible for some of the differences observed. CONCLUSION: Compared to other statin users without recent prior cardiovascular events, the incidence of fatal and non-fatal CVD in thisretrospective observational cohort study was 28% lower among rosuvastatin users.