Controversy surrounds which normative data should be used to estimate osteoporosis prevalence in men. Prevalence estimates may vary significantly when different normative standards are applied. Five normative datasets (NHANES female norms, local female norms, Hologic densitometer manufacturer female norms, NHANES male norms, Hologic male norms) were used to estimate the prevalence of osteoporosis by World Health Organization diagnostic criteria in a study population of 311 consecutive men between the age of 30 and 91 (mean 60.3 yr) referred to an outpatient osteoporosis center between January 1996 and December 1998. Statistically significant variations were seen in osteoporosis prevalence measured at three anatomical sites. The greatest relative variation was seen for the total femur, where osteoporosis prevalence ranged from 7.0% (NHANES and Hologic female norms) to 15.6% (NHANES male norms). The least relative variation was seen at the lumbar spine, where prevalence ranged from 18.1% (Hologic female norms) to 29.6% (local female norms). When considering osteoporosis at any site, prevalence was lowest (23.5%) based on Hologic female norms and highest (35.8%) based on local female norms. Interpretation of prevalence data should include an assessment of how normative standards influence reporting of the population at high risk of fracture.