Hospitals in France are encouraged to monitor antibiotic consumption (AbC) and it is known that this differs among hospitals. The aim of the current study was to identify relevant and easily available adjustment criteria for the purpose of benchmarking. We analysed data from 34 public non-teaching hospitals and 43 private hospitals located in south-western France and overseas departments using retrospective data from 2005. This study investigated the relationship between AbC expressed as defined daily doses per 1000 patient-days (DDD/1000 PDs) or per 100 admissions (DDD/100 admissions) and the number of venous central lines, the number of episodes of bacteraemia and various hospital characteristics. The relationship was tested using multiple linear analyses. The median total AbC in public hospitals was 395 DDD/1000 PDs (range, 196-737) and 341 DDD/100 admissions (range, 180-792). In private hospitals this was 422 DDD/1000 PDs (range, 113-717) and 212 DDD/100 admissions (range, 38-510). The best model for public hospitals included the proportion of PDs in surgery, intensive care and medical wards and explained 84% of the variability in AbC expressed as DDD/1000 PDs. For private hospitals, the mean length of stay and the proportion of PDs in surgery and medical wards explained 68% of the variability in AbC expressed as DDD/100 admissions. Overall, this French experience shows that relevant adjustment criteria for the comparison among hospitals are easily available. It is important that each country establish its own model considering the intrinsic peculiarities of the hospital system and taking into account both indicators (DDD/1000 PDs or DDD/100 admissions) to design the best model.