The aim of this study was to compare the muscle grade and bulk of the divided and repaired piriformis tendon by the standard posterior approach with the preserved tendon by a piriformis-sparing (PS) approach, in total hip arthroplasty. Twenty-two patients were randomized to either approach. Patients received preoperative and 3 months and 2 years postoperative magnetic resonance imaging scans. Patients and evaluators were blinded to allocation. There was significantly less deterioration in piriformis muscle grade (P = .029) and bulk (P = .015) in the PS group at 3 months. At 2 years, only the difference in grade remained significant (P = .001). There was no difference in Oxford hip scores. In conclusion, a PS approach avoids the marked wasting and deterioration in muscle grade that occurs postoperatively in the standard posterior approach.