Average femoral geometry was determined from sections of 25 human femurs. A 'stem design program' was used to determine the largest femoral stem that could be introduced into the canal. The stem was fine tuned by observations of fit in Plexiglass models of the 25 femurs. This stem was called the 'Exact-Fit'. Type 1 stem was formed by grinding the anterior and posterior faces flat. Type 2 stem was formed by additionally grinding the medial and lateral faces flat. Four fresh femurs were used for the experiments. Strain gages were located on the proximal medial area and level with the distal lateral tip. Displacement transducers were mounted proximally to measure stem-bone micromotion. Vertical and anterior-posterior forces were applied to the femoral head and normal strain data obtained. The stems were then inserted in sequence and the strains and micromotions were measured. After the uncemented stems had been tested, a smaller size of 'exact-fit' stem was cemented and the tests repeated. The strains with the uncemented stems were closer to normal than with the cemented stems. For example, compressive calcar strains averaged 56% of normal uncemented compared with 30% cemented for the different loadings. The micromotion was higher uncemented, but was still at a low level. The 'exact-fit' stem gave less resultant micromotion than the type 1 and type 2 stems. The high variation of the strain results was attributed to variations in stem-bone contact and fit. The study provided information of direct application to press-fit stems in hip replacement surgery.