Anti-remodeling agents produce similar reductions in vertebral fracture risk despite large differences in BMD changes suggesting the mechanism of fracture risk reduction may differ among these agents. Forty-eight intact (non-ovariectomized) skeletally mature female beagle dogs were treated orally for 12 months with clinically relevant doses of risedronate (RIS, 0.10 mg/kg/day), alendronate (ALN, 0.2 mg/kg/day), raloxifene (RAL, 0.50 mg/kg/day), or saline (VEH, 1 ml/kg/day). After sacrifice, the following measurements were made on vertebral bone: areal (aBMD) and volumetric (vBMD) bone mineral densities, tissue mineralization by ash content, static and dynamic histomorphometric parameters, microdamage, and extrinsic and intrinsic measures of biomechanical strength, stiffness and energy to fracture. At these doses, RAL suppressed bone turnover (-20%) significantly less than the bisphosphonates (-66 and -71%) and did not produce significant differences in aBMD, vBMD, BV/TV or percent ash compared to VEH-treated animals. Microdamage accumulation in RAL-treated animals was not significantly different than VEH; both RIS and ALN had significantly higher crack surface density compared to VEH. Stiffness was significantly higher than VEH in all treatment groups. Ultimate load divided by aBMD, a measure of strength independent of BMD, was significantly higher only in RAL-treated animals compared to VEH (+16%, P = 0.015). Based on these data, we conclude that raloxifene produces improvements in bone mechanical properties in ways that do not involve increases in BMD.