Densitometric measurements are prone to imprecision in elderly subjects and the present study was primarily designed to dissect out the effects of age and bone mineral density on proximal femur dual energy x-ray absorptiometry (DXA) reproducibility. The study comprised 17 elderly women (mean age 74.6 years, range 65-84 years), 13 early postmenopausal women with osteopenia (mean age 56.2 years, range 50-63 years), and 17 elderly men (mean age 73.8 years, range 65-86 years). Each subject was given triplicate proximal femur scans by a QDR 2000 Densitometer (Hologic Inc., Waltham, MA) with repositioning between scans. Because of subject selection in the early postmenopausal women there were no significant differences in bone mineral density (BMD) at any site among the three groups. Despite this, reproducibility errors expressed as either coefficient of variation (CV) % or mean standard deviation (SD) were greater in the elderly subjects, regardless of gender, when compared with the younger female subjects. The variability in measurement errors with age were least marked for the total hip and trochanteric sites. Within the elderly subjects, BMD appeared to exert little influence on measurement errors. We conclude that short-term proximal femur reproducibility is dependent on age-related factors other than BMD. There is no influence of gender on the measurement errors. It is likely that local factors (e.g., hip osteoarthritis) or general frailty may influence repositioning but this needs further exploration. In the meantime, the total hip and trochanteric sites should be used as they provide the most reproducible measurements in the elderly.