Hip axis length (HAL) has been reported as an independent risk factor for hip fracture. DEXA machines using fan beam techniques have become increasingly available. Errors in calculated hip axis length may be expected because of different degrees of magnification by the fan beam. The magnitude of this error on measurement of hip geometry was studied, using an anthropomorphic femur phantom with both fan beam (Lunar Expert and Hologic QDR-2000) and pencil beam (Lunar DPXL) densitometers. The clinical relevance of these findings was also examined using patient measurements of buttock soft tissue thickness. Femoral neck axis length (FNAL), which correlates well with HAL, was used as a measurement of hip geometry. There was a linear increase or decrease of FNAL with increasing distance of the phantom above the scanning table, when measured with the Lunar Expert or Hologic QDR-2000, respectively. There was no significant difference in FNAL at different heights using the pencil beam densitometer. The maximal difference in buttock soft tissue thickness in 30 women studied was 8.7 cm. From the equations, derived from the phantom studies, this difference would result in an 8.2% (1.4 SD) increase, or an 11.4% (1.9 SD) decrease in FNAL in the largest woman as compared with the smallest woman when measured using the Lunar Expert or Hologic QDR-2000, respectively. We conclude that there may be unpredictable degrees of magnification of FNAL in vivo, caused by differences in buttock thickness, when measured using a fan beam densitometers. Until these problems are resolved. FNAL, or related parameters of femoral geometry, should be measured using pencil beam instruments.