Between 1987 and 2006 we performed a modified Thompson's quadricepsplasty on 40 fracture-related stiff knees and followed the patients for a mean of 7.9 years (2 to 11.1). The factors affecting the final gain of movement were investigated. A total of 15 knees required lengthening of the rectus femoris. The mean flexion gain was 70.2° (42.3° to 112.5°). According to Judet's criteria, the results were excellent in 30 knees, good in seven, and fair in three. The range of movement which was achieved intra-operatively was related to the gain of knee flexion on univariate analysis. Five patients had complications: deep infection in one, recurrent patellar dislocation in one, and rupture of the extensor mechanism in three. This modified technique gives satisfactory results. Achieving maximum knee flexion intraoperatively seems to be the most important factor in enhancing the outcome in patients with stiffness of the knee following fracture.