Aims: The concept of balancing ßexion and extension gap during total knee arthroplasty (TKA) is reported to be crucial. However, difþculties in 1)deþning the ideal distraction force to create a gap, and 2)equalizing ßexion and extension gap are often encountered during TKA for rheumatoid arthritis (RA). This study was performed to analyze these difþculties biomechanically in vivo. Methods: 35 knee at randomly selected TKA for RA were studied as follows. After soft tissue balancing, distraction force for both gaps was applied by ligament balancer. Force was gradually increased with recording the length of the gap created by consecutive force, in order to obtain load-elongation curve for each case. Results: Load-elongation curve showed various patterns, indicating soft tissues including ligamentous structure has been altered its biomechanical property due to the variety of rheumatoid pathology. If ideal force for extension gap was determined at the point when low stiffness changes to high stiffness in the curve, it would be about 120–200N. This force differed reasonably in each case. However, measured ßexion gap curve hardly reached this force in more than 60% of the35 knee, presumably due to functional loss of posterior structures. Interestingly, this phenomenon was often unpredictable at examination before the operation. Conclusions: Ideal tension for þlling the gap with implant seems to be different in each case from load-elongation curve analysis. From this study, we raise question to the concept of equal ßexion and extension gap. This concept, although sounds attractive, is often difþcult to obtain in rheumatoid knee. This observation may suggest which type of TKA (þxed or mobile) is safer for replacing the rheumatoid knee.