Twelve normal greyhound femora were divided into three groups. In group one, femoral stems were placed in neutral position with maximal fill. Group two had undersized femoral stems placed in neutral position. Group three had undersized femoral stems placed in varus position. Intact and implanted femora were loaded from 10 newtons (N) to 300 N in axial compression at a rate of 25 N/s for 10 replications. A strain gauge analysis showed that the strain distribution of all implanted femora were substantially different from intact femora, but femora with large implants placed in neutral position had the least amount of deviation from normal. An undersized stem in neutral position had significantly less compressive longitudinal strains along the proximomedial and proximocranial cortices. An undersized stem in varus position improved implant fit along the proximomedial and distolateral cortices, which resulted in increased tensile hoop strains. There were multiple significant correlations between the strain data and implantation variables (implant alignment, canal fill, and implant fit). Subsidence was significantly greater for the undersized implant in neutral position. There was not a difference in subsidence between the large neutral and varus groups. The most important variable that decreased subsidence was increased lateral implant fit (r= -0.86, P= .0003).