The effect of 5 fracture fixation methods on bone mineral density (BMD) measurement of femurs, using dual-energy X-ray absorptiometry (DXA) was determined in a canine model. Six regions of interest were measured, including the entire femur, the diaphysis of the femur, and small regions centered over the middiaphysis of the bone (lateral middiaphyseal cortex, medial middiaphyseal cortex, middiaphyseal medullary canal, and total middiaphysis). Eight unpaired femurs were collected and scanned by use of DXA before (5 separate scans/femur) and after (5 separate scans/femur) fixation by use of 1 of 5 fixation methods. These fixation methods included: intramedullary (IM) nail: IM nail and cerclage wires; IM nail and external skeletal fixation.; locked IM nail; and a dynamic compression plate (DCP). All implants were made of stainless steel. The IM nail fixation devices caused significant decreases in the DXA measurement of BMD in the small regions of interest, compared with femurs without fixation devices (mean decrease, 37.3%; P < 0.05). The locked nail caused similar, but larger, decreases in the DXA measurement of BMD, compared with the IM nail fixation methods (P < 0.05). Plate fixation caused a small, but significant (P < 0.05), decrease (2.8%) in the DXA measurement of BMD in the large regions of interest, but when all regions were averaged, it did not cause significant change in this measurement, compared with femurs without fixation devices. Plate fixation caused a large change in the DXA measurement of BMD in 1 region only in the lateral cortical bone under the plate where the DXA measurement of BMD was increased 13.3% over that in femurs without fixation devices (P < 0.05). In femurs without fixation devices, the precision error ranged from 0.5% for large regions of interest to 2.4% for small regions of interest. None of the fixation methods altered the precision error of large regions of interest (P > 0.05). In contrast, the precision errors of the small regions of interest were increased by the IM fixation methods and the locked IM nail, When all regions were combined, IM fixation methods caused significant (P < 0.05) increase in the precision error, compared with femurs without fixation devices (mean increase, 157%; range, 121 to 193%). Plate fixation did not change the precision error of any region of interest, compared with femurs without fixation devices (P > 0.05; power = 0.8 at delta = 64%).