Accurate knee morphology is of value in determining the correct sizing of prosthetic implants. Intraoperative measurement of key linear dimensional variables was carried out on 196 Caucasian knees (osteoarthritic patients: 68 male and 128 female). Of the 196 knees measured, 70 had extensive cartilage degeneration. Statistical analysis was carried out on this large sample size of data. Summary statistics and correlation coefficients between variables were determined and compared between subgroups. Male knees were on average larger than female knees. Higher correlation was found between variables for males than between variables for females. Overall, the patellar dimensions were seen to correlate least well with other anatomical variables. High correlation between femoral variables supports current femoral sizing procedure, although routine patellar resection practices are called into question. Average values for the 70 knees with extensive cartilage degeneration were significantly smaller (P < 0.01) than their counterparts for the other 126 knees. For a measurement not containing cartilage, such as femoral epicondylar width, this difference cannot be accounted for by the loss of cartilage owing to wear. This suggests that, for similar height and weight, a naturally narrower femoral epicondylar width may be associated with severe osteoarthritis.