This study was designed to determine which subscapularis repair method performs best under fatigue loading. Fresh-frozen human cadaveric shoulders were used. A standard circumferential release was performed, including a rotator interval release and separation of the subscapularis from the anterior and inferior capsules. One of 3 methods was used to repair the subscapularis tendon: tendon-to-tendon, tendon-to-bone, or bone-to-bone using a buttress plate. The shoulder was mounted on a fatigue-testing machine, and the medial end of the subscapularis tendon was attached to a soft-tissue clamp. We initially tested each specimen for fatigue at 150 N for 500 cycles, then at 300 N for 2500 cycles. The failure rate of bone-to-bone and tendon-to-tendon repairs was significantly better than that of the tendon-to-bone repair. The bone-to-bone repair exhibited the best combination of repair strength and restoration of subscapularis length.