Cementless femoral stems were placed into 12 normal greyhound femora. The implanted femora were divided into three groups by stem orientation and implant size and loaded in axial compression at a rate of 25 newtons (N) per second until failure. Rosette strain gauges were used to measure femoral principal strains at 500 N, 1,000 N, 1,500 N, and at maximum load. During maximum load, varus orientation of the femoral stem had significantly higher tensile hoop strains in the proximomedial cortex, whereas neutral orientation had higher tensile hoop strains along the cranial cortex. Femoral fractures occurred in these areas of peak tensile strain. There was no difference in maximum load between groups, therefore varus orientation did not predispose to fracture. Maximizing canal fill and implant fit increased implant stability.