The influence of subject repositioning on the precision of bone mineral densitometry of the proximal femur using dual energy X-ray absorptiometry was estimated by determining the variation in the bone mineral density (BMD) of subjects scanned with their foot and leg position varying from that routinely used in clinical scanning. The mean variations in BMD of the femoral neck, Ward's triangle and trochanter were 2.7, 4.1 and 1.7%, respectively, in eight subjects scanned with the foot internally rotated by 0, 13 and 27 degrees. The mean variations in BMD in four subjects scanned with the leg in the customary position and abducted +/- 6 degrees from the conventionally used scanning position were 3.6, 2.8 and 1.6% for the same respective regions of interest. For diagnostic applications, the orientation of the leg and foot during scanning is relatively unimportant since the variations in BMD introduced by the different foot and leg orientations likely to be encountered in routine clinical scanning are small compared to the intrapopulation variation in BMD. However, for monitoring changes in BMD longitudinally, careful repositioning of the foot and leg will be necessary to achieve precision in vivo of less than 1%.