Finite element (FE) models could be used for pre-clinical testing of cemented hip replacement implants against the damage accumulation failure scenario. To accurately predict mechanical failure, the models should accurately predict stresses and strains. This should be the case for various implants. In the current study, two FE models of composite hip reconstructions with two different implants were validated relative to experimental bone and cement strains. The objective was an overall agreement within 10% between experimental and FE strains. Two stem types with different clinical results were analyzed: the Lubinus SPII and the Mueller Curved with loosening rates of 4% and 16% after 10 yr, respectively (Prognosis of total hip replacement. 63rd Annual Meeting of the American Academy of orthopaedic surgeons, Atlanta, USA). For both implant types, six stems were implanted in composite femurs. All specimens were subjected to bending. The Mueller Curved specimens were additionally subjected to torsion. Bone strains were recorded at 10 locations on the cortex and cement strains at three locations within the cement mantle. An FE model was built for both stem types and the experiments were simulated. Bone and cement strains were calculated at the experimental gauge locations. Most FE bone strains corresponded to the mean experimental strains within two standard deviations; most FE cement strains within one standard deviation. Linear regression between the FE and mean experimental strains produced slopes between 0.82 and 1.03, and R/sup 2/ values above 0.98. Particularly for the Mueller Curved, agreement improved considerably when FE strains were compared to the strains from the experimental specimen used to build the FE model. The objective of overall agreement within 10% was achieved, indicating that both FE models were successfully validated. This prerequisite for accurately predicting long-term failure has been satisfied.