A two-dimensional finite element model was used to compare the femoral head stress distribution of an intact femur with that of a femur following core biopsy and biopsy with cortical grafting. There was a decrease in superior and lateral femoral head stresses of up to 50% by an incorporated and properly placed cortical graft. Core biopsy and cortical grafting short of the subchondral plate produced stress-concentrating effects. A medially placed cortical graft did not stress-relieve the superolateral femoral head. Thus, if cortical grafting is to be undertaken, the graft should extend from the lateral cortex to the subchondral plate, transfixing the length of the necrotic segment. Because of deleterious effects on femoral head stress distribution, core biopsy without cortical grafting should be discouraged or only very cautiously used.