Background: Hospital antibiotic use can be measured by calculating daily doses as defined by the WHO/ATC index (DDD) divided by the number of patient or occupied bed days. We wondered whether changes in antibiotic use density over time at a university hospital using this data format are similar in order of magnitude when compared with a different, alternative dose definition and the number of admissions rather than the number of patient days as denominator. Methods: Data obtained from the hospital pharmacy for the medical and surgical services of a 1,000-bed university hospital for the period 1992 through 2003 were expressed both in daily doses per 100 patient days and daily doses per admission. A PDD dose definition (prescribed daily doses), defining doses that reflect the usually prescribed dose in adult hospitalized patients with normal renal function was compared with the WHO/ATC 2001 DDD dose definitions. The percent changes using the different data formats between two 3-year averages (1992-1994 and 2001-2003) were calculated. Results and Conclusion: The DDD/100 patient days data format overestimated antibiotic use density changes in this hospital both in medicine (81% vs 48%) as well as in surgery (69% vs 39%) when compared with PDD/100 patient days. Due to changes in the number of admissions and length of stay over the years, the percent change between the two periods expressed in doses per 100 patient days in addition differed substantially from that estimated by using the DDD per admission or PDD per admission data format. Studies evaluating the evolution of hospital antibiotic use need to address the limitations and adequacy of the different data formats.