AIMS: The defined daily doses (DDD) defined by the WHO are widely used as an indicator to measure antibiotic use in the hospital setting. However, discrepancies exist between countries in terms of antibiotic dosage. The aim of the present study was to compare, for each antibacterial agent available at our university hospital, the prescribed daily doses (PDD) with the DDD. METHODS: Data were extracted from the pharmacy computer system. Antibiotic use was expressed in DDD per 1000 patient days. We also calculated the ratio of number of DDD:number of treatment-days and estimated the average PDD for each antibiotic and route of administration. RESULTS: The average PDD did not correspond to the DDD for many classes of antibiotics. If fluoroquinolones and cephalosporins were prescribed at a dosage close to the DDD, other antimicrobial classes such as penicillins, aminoglycosides or macrolides were not. Overall, the number of DDD overestimated the number of treatment days by 40%. For the most consumed antibiotic at our hospital, i.e. oral amoxicillin-clavulanic acid, the PDD was three times the DDD. CONCLUSIONS: Our study shows that, except for the fluoroquinolones and the cephalosporins, the number of DDD did not correctly reflect the number of antibiotic treatment days at our hospital. This does not invalidate the systematic approach of the WHO and hospitals should use the DDDs to make national and international comparisons of their antibiotic use. However, each hospital should define and validate its own indicators to describe the local exposures to antibiotics and to study the relationship with resistance.