BACKGROUND Implants are commonly used to stabilize proximal femoral osteotomies in children with cerebral palsy (CP). Removal of implants is common practice and believed to avoid infection, fracture, or pain that might be associated with retained hardware. There is little evidence to support a prophylactic strategy over a reactive approach based on symptoms. The aim of this study was to compare the outcomes of prophylactic and reactive approaches to removal of proximal femoral implants in children with CP. METHODS An intention-to-treat model was used to compare 2 institutions that followed a prophylactic (within ∼1 y) and reactive (following complication/symptoms) approach to hardware removal, respectively. Patients with CP who had femoral implants placed at or before age 16, and had ≥2-year postsurgical follow-up were included. Demographics, surgical details, reasons for removal, and complications were recorded. χ and t tests were used. RESULTS Six hundred twenty-one patients (prophylactic=302, reactive=319) were followed for an average of 6 years (range, 2 to 17 y). Two hundred eighty-seven (95%) implants were removed in the prophylactic group at 1.2 years. In the reactive group, 64 (20%) implants were removed at an average of 4.2 years. Reasons for removal included pain; infection; fracture; or for repeat reconstruction. The rate of unplanned removals due to fracture or infection was higher in the reactive group (4.7% vs. 0.7%, P=0.002), but there was no difference in the rate of complications during/after removal between the 2 groups (1.7% vs. 3.1%; P=0.616). No specific risk factor associated with unplanned removal could be identified; but children under 8 years old seemed more likely to undergo later removal (odds ratio 1.98; 95% confidence interval, 0.99-3.99). CONCLUSIONS Eighty percent of patients in the reactive removal strategy avoided surgery. This group did have a 4% higher rate of fracture or infection necessitating unplanned removal but these were successfully treated at time of removal with no difference in complication rates associated with removal between both groups. One would need to remove implants from 25 patients to avoid 1 additional complication, providing some support for a reactive approach to removal of proximal femoral implants in this population. METHODS Level III-therapeutic.