Osteoporosis is common among the growing population of older men: almost 20% of men > or = 50 years old have osteoporosis of the hip, spine, or wrist. However, the exact estimate depends on the approach taken to normalize for bone size, the specific skeletal site assessed, and the diagnostic criteria used. Bone mineral density (BMD, g/cm2) by DXA is 12-25% greater in men than women, but bone mineral apparent density (g/cm3) is similar in the two sexes. This correction for skeletal size largely eliminates apparent differences in areal BMD between the races and also reduces the apparent effects on BMD of age. The particular skeletal site that is assessed has an important influence on the prevalence of osteoporosis (sex-specific BMD T-score less than -2.5) in men which varies from 0 to 36%, depending on the site, and from 2% to 45% in postmenopausal women. The discrepancies relate mainly to different patterns of bone loss at the various sites, but estimates are also affected by the specific young normal means and standard deviations (SD) used to calculate the T-scores. A greater mean and smaller SD among normal young men in Rochester, MN produced a higher prevalence of osteoporosis at the femoral neck (22% vs 7%) compared with estimates for white men from the Third National Health and Nutritional Examination Survey; use of female normal values further reduced osteoporosis prevalence at the hip in white, Hispanic, and African-American men to 4%, 2%, and 3%, respectively, compared with 20% for white women in the United States. By contrast, fracture risk is similar for men and women at any given level of BMD. These observations reinforce current efforts to move away from osteoporosis prevalence and toward absolute fracture risk as the main basis for clinical treatment decisions.