As part of a prospective study of 476 total knee replacements (TKR), we evaluated the use of manipulation under anaesthesia in 47 knees. Manipulation was considered when intensive physiotherapy failed to increase flexion to more than 80°. The mean time from arthroplasty to manipulation was 11.3 weeks (median 9, range 2 to 41). The mean active flexion before manipulation was 62° (35 to 80). One year later the mean gain was 33° (Wilcoxon signed-rank test, range −5 to 70, 95% CI 28.5 to 38.5). Definite sustained gains in flexion were achieved even when manipulation was performed four or more months after arthroplasty (paired t-test, p < 0.01, CI 8.4 to 31.4).

A further 21 patients who met our criteria for manipulation declined the procedure. Despite continued physiotherapy, there was no significant increase in flexion in their knees. Six weeks to one year after TKR, the mean change was 3.1° (paired t-test, p = 0.23, CI −8.1 to +2).