The aim of this study was to determine whether the dimensions of the distal femur and proximal tibia joint surfaces affect the etiology of knee osteoarthritis (OA). The study comprised the records of 1,324 patients who had been admitted to hospital with knee pain. Anterioposterior (AP) and lateral radiographs of the knee were taken. Using the Kellgren-Lawrence Scale, the patient group comprised Stages 2, 3, and 4 radiographs and the controls comprised Stages 0 and 1 radiographs. Four lengths were measured for each patient in both groups: femur mediolateral (femur ML), tibia mediolateral (tibia ML), femur anteroposterior (femur AP), and tibia anteroposterior (tibia AP). Osteophytes were not included in the measurements in the patient group. All the measurements were repeated by two researchers at two different times. The groups were compared in terms of these measurements and the correlations between them. The mean femur ML length was significantly greater in the patient group than the control group (P=0.032) and the mean femur AP length was significantly less (P=0.037). In addition, the difference between the femur ML and AP lengths was significantly high in the patient group (P<0.001). The difference between the tibia and femur ML lengths was significantly high in the patient group (P<0.001) and the difference between the tibia and femur AP lengths was higher in the control group (P=0.001). A longer femur ML and a shorter femur AP, together with a greater difference between these two lengths and a greater difference between the tibia ML and femur ML lengths, could be a risk factor for developing knee OA. More extensive anatomical and biomechanical studies in the future will enable these results to be corroborated. Clin. Anat. 28:672-677, 2015. (c) 2015 Wiley Periodicals, Inc.