The ability of bone to adapt to its mechanical environment is well recognized, although the specific mechanical parameters initiating or maintaining the adaptive responses have yet to be identified. Recently introduced mathematical models offer the potential to aid in the identification of such parameters, although these models have not been well validated experimentally or clinically. We formulated a complementary experimental/analytic approach, using an animal model with a well-controlled mechanical environment combined with finite element modeling (FEM). We selected the functionally isolated turkey ulna, since the loading could be completely characterized and the periosteal adaptive responses subsequently monitored and quantified after four and eight weeks of loading. Known loads input into a three-dimensional, linearly elastic FEM of the ulna then permitted full-field mechanical characterization of the ulna. The FEM was validated against a normal strain-gaged turkey ulna, loaded in vivo in an identical fashion to the experimental ulnae. Twenty-four candidate mechanical parameters were then compared to the quantified adaptive responses, using statistical techniques. The data supported strain energy density, longitudinal shear stress, and tensile principal stress/strain as the mechanical parameters most likely related to the initiation of the remodeling response. Model predictions can now suggest new experiments, against which the predictions can be supported or falsified.