Repeat spine and femur measurements (5 per case) were done on 19 subjects with the DPX-L densitometer operating at 3 mA giving a radiation flux fourfold higher than the earlier DPX model. The precision for spine bone mineral density (BMD) was about 0.55% (L2-L4) and 0.48% (L1-L4) for 2-minute scans (2.4 mrem). The precision was only slightly lower (0.4-0.5%) for 4-minute scans (5 mrem) in a subset of 11 subjects. There was a slight precision advantage for the larger L1-L4 area compared with L2-L4 for 2-minute scans, but no advantage for 4-minute scans. The precision for femoral neck BMD was 1.00 and 0.85% for 2- and 4-minute scans, respectively, with proper positioning. The corresponding values for the Ward's triangle region of the femur were 2.6 and 1.5%. The precision of spine scans was influenced chiefly by variable region location. The precision of femur scans was affected by both patient positioning and location of the region. The 4-minute scans minimized the number of operator changes necessary for analysis. Precision errors can be reduced by up to 50% with utilization of the higher flux, but this does not obviate the need for care in patient positioning and scan analysis.