The aim of this study was to evaluate the influence of the Gross Motor Function Classification System (GMFCS) on the outcomes of rectus femoris transfer (RFT) for patients with cerebral palsy and stiff knee gait. We performed a retrospective review of patients seen at our gait laboratory from 1996 to 2013. Inclusion criteria were (i) spastic diplegic cerebral palsy, (ii) GMFCS levels I-III, (iii) reduced peak knee flexion in swing (PKFSw<55°), and (iv) patients who underwent orthopedic surgery with preoperative and postoperative gait analysis. Patients were divided into two groups according to whether they received a concurrent RFT or not at the time of surgery: non-RFT group (185 knees) and RFT group (123 knees). The primary outcome was the overall knee range of motion (KROM) derived from gait kinematics. The secondary outcomes were the PKFSw and the time of peak knee flexion in swing (tPKFSw). We observed a statistically significant improvement in KROM only for patients in the RFT group (P<0.001). However, PKFSw and tPKFSw improved in both groups after surgery (P<0.001 for all analyses). In the RFT group, the improvement in KROM was observed only for patients classified as GMFCS levels I and II. In the non-RFT group, no improvement in KROM was observed in any GMFCS level. In this study, patients at GMFCS levels I and II were more likely to benefit from the RFT procedure.