As more therapies are introduced to treat osteoporosis, precise in vivo methods are needed to monitor response to therapy and to estimate the gains in bone strength that result from treatment. A method for evaluating the strength of the proximal femur was developed and its short term reproducibility, or precision, was determined in vivo. Ten volunteer subjects aged 51–62 years (mean 55.6 years), eight women and two men, were examined using a quantitative computed tomography (QCT) protocol. They were positioned, scanned, re-positioned and re-scanned. The QCT images were registered in three-dimensional space, and finite element (FE) models were generated and processed to simulate a stance phase load configuration. Stiffness was computed from each FE model, and strength was computed using a regression equation between FE stiffness and fracture load for a small set n=6 of experimental specimens. The coefficients of variation (COV) and repeatability (COR=2.23* √2*COV) were determined. The COV for the FE fracture load computed was 1.85%, and the detectable limit (coefficient of repeatability) for serial measurements was 5.85%. That is, if a change of 5.85% or more in computed FE fracture load is observed, it will be too large to be consistent with measurement variation, but instead can be interpreted as a real change in the strength of the bone. The detectable limit of this method makes it suitable for serial research studies on changes in femoral bone strength in vivo.