Hospital doctors influence general practitioners’ prescribing patterns. But, no study has focused on the influence of hospital choices of drugs on prescribing patterns in the community for competitive pharmacological classes at the macro level. OBJECTIVES: To investigate the influence of hospital choices of drugs on the pharmaceutical consumption in its surrounding community. METHODS: Quasi-experimental study. Data from the national survey about drugs in hospitals of the French Department of Health were used to extract purchased quantities in 2008 by 25 of the 29 French university hospitals for proton pump inhibitors (PPI), serotonin antagonists (setrons), low molecular weight heparins (LMWH), erythropoietins (EPO), angiotensin conversion enzyme inhibitors (ACEI), angiotensin II receptors antagonists (ARA), statins, alpha-adrenoreceptor antagonists (AAA) and selective serotonin reuptake inhibitors (SSRI). Reimbursed quantities from general practitioners’ prescriptions for patients living in the hospital’s catchment area were extracted from the national health insurance database (concerning 90% of the population). Quantities for each brand and each drug entity were expressed in defined daily doses per 1000 hospitalizations-day (DHD) or 1000 inhabitants-day (DID) and log-transformed. Correlations between consumption in university hospitals and their catchment areas were assessed using Spearman rank correlation coefficients. RESULTS: University hospitals selected between 30 and 52% of the available drug entities for these 9 classes. A correlation coefficient superior to 0.20 was calculated for 45% of brand names and 75% of the drug entities. Significant correlations were found for the classes of SSRI and ACEI. For statins, ARA, LMWH and setrons, correlation was quite good but did not achieve statistical significance boundary. For the remaining classes (PPI, EPO and AAA), correlation varied with the drugs, but was globally nonexistent. CONCLUSIONS: Correlations between hospital and community uses varied with classes and drugs. The direction of the relationship, as well as its determinants, will be further assessed in a multivariate analysis.