Although a substantial amount is known about the pathogenesis of osteoarthrosis, its aetiology and in particular the role that mechanical factors play, remains unclear. One particular hypothesis suggests that cartilage adapts mechanically so that it may transmit, without sustaining damage, the stresses to which it is predominantly subjected, and that damage to the cartilage is caused by infrequent high stresses in excess of the predominant level. As a corollary, it was suggested that highly stressed cartilage should be stiffer than lowly stressed cartilage. A survey of the mechanical properties of normal articular cartilage from unembalmed cadaveric knee and ankle joints was undertaken to test this hypothesis. For this purpose, a specially developed indentation test apparatus was commissioned. Tests of the machine's measurement capabilities indicate that coefficients of variation of 2.14% and 1.20% for indentation and cartilage thickness measurement could be expected. The maximum percentage errors in the calculated creep modulus value which could result from these typical measurement errors, were 4.2% and 2.9% respectively. Creep modulus values, calculated from these measurements, were used in topographical comparisons of cartilage stiffness. The stiffest areas of cartilage in the knee joint were the femoral condyles and areas of the tibia covered by the menisci. Cartilage on the patellar surfaces of the femur and in areas exposed by the menisci was significantly softer. Cartilage from the ankle joint was considerably stiffer than cartilage from the knee. Comparisons between the cartilage stiffness and levels of stress which act in the knee and ankle joints during normal ambulatory activity, showed the stiffest areas of cartilage to be subjected to the greater stresses. Correlations of averaged data values indicated a significant (p < 0.01) direct relationship between cartilage stiffness and stress. This relationship and the consistency with which osteoarthrotic lesions were found in areas subjected to damaging patterns of stress supported the hypothesis under examination. The lack of correlation found between the proteoglycan content and cartilage stiffness suggested that structural rather than compositional factors may be more important in influencing the compressive stiffness of normal articular cartilage.