Aims: Three methods of mobilization are currently performed: arthroscopic release (AR), manipulation under general anesthesia (MUGA), open surgical release (OSR). This study assessed the accurate indications of these 3 procedures to treat stiff knee arthroplasties. Methods: Sixty-two of these procedures were performed between 1989 and 2001 and followed at least 1 year: 34 MUGA, 18 AR and 10 OSR. The 3 groups were comparable excepted for the delay between the prosthetic insertion and the mobilization procedure: 17 weeks for MUGA, 46 weeks for AR, 97 weeks for OSR. All the patients had the same postoperative analgesia and rehabilitation program. Results: For the 62 procedures there was an improvement in range of ßexion from preop-erative (mean 58.4¡) to follow-up (mean 94.6¡) and a decrease in ßessum deformity from 7.6¡ to 2.5¡ (p= 0.001). From surgery to 1 year of follow-up, there was a decrease in ßexion (104.6¡ to 94.6¡) and an increase in ßessum deformity (1.3¡ to 2.5¡) (NS). The worst postoperative ranges of motion were observed at 6 weeks after the procedure, and then an improvement was observed up to 6 months but was non-signiþcant. Flexion did not improved beyond 6 months after the procedure. The results of the 3 techniques were not signiþcantly different. However, failures were more frequent when MUGA were performed beyond 8 weeks after prosthetic insertion, and when AR were performed beyond 6 months after prosthetic insertion (p< 0.01). Conclusions: We recommend to treat stiff total knee prostheses by MUGA until 8 weeks after insertion, by AR between 8 weeks and 6 months, and by OSR later on. This protocol addresses stiff prostheses without infection and without component malposition. The deþnitive ranges of motions were obtained at 6 months after mobilization.