Accurate computer modelling of the fixation of unicompartmental knee replacements (UKRs) is a valuable design tool. However, models must be validated with in vitro mechanical tests to have confidence in the results. Ten fresh-frozen cadaveric knees with differing bone densities were CT-scanned to obtain geometry and bone density data, then implanted with cementless medial Oxford UKRs by an orthopaedic surgeon. Five strain gauge rosettes were attached to the tibia and femur of each knee and the bone constructs were mechanically tested. They were re-tested following implanting the cemented versions of the implants. Finite element models of four UKR tibiae and femora were developed. Sensitivity assessments and convergence studies were conducted to optimise modelling parameters. The cemented UKR pooled R(2) values for predicted versus measured bone strains were 0.85 and 0.92 for the tibia and femur respectively. The cementless UKR pooled R(2) values were slightly lower at 0.62 and 0.73 which may have been due to the irregularity of bone resections. The correlation of the results was attributed partly to the improved material property prediction method used in this project. This study is the first to validate multiple UKR tibiae and femora for bone strain across a range of specimen bone densities.