Factors contributing to subsidence were analyzed by radiographic evaluation and mechanical testing of 36 canine cadaver femora during and after insertion of an uncemented porous-coated femoral stem and by radiographic evaluation of 35 canine total hip arthroplasties. Mean percentage of canal fill in immediate postoperative radiographs, and percentage of canal fill at midimplant and distal implant locations, were accurate predictors of subsidence. Force required to implant the femoral stem was strongly correlated with force required for implant subsidence. Femoral morphology and percentage of canal fill at the middle and distal sites were accurate predictors of subsidence. Implants in femora with a stovepipe morphology (canal flare index ≤ 1.8) were six times more likely to subside than implants in femora that had a normal appearance (canal flare index 1.8 to 2.5), and 72 times more likely to subside than implants in champagne-fluted femora (canal flare index ≥ 2.5). Femora with more than 85% mean, middle, or distal canal fill were less likely to subside.