Vascularized tendon grafts were compared with nonvascularized tendon grafts in a primate experimental model. In four monkeys, seven vascularized extensor hallucis longus grafts were placed in the foot's digital fibroosseous canal and these were compared with eight nonvascularized tendon grafts similarly placed in the opposite extremities. The juncture techniques and postoperative protocols were identical for both tendon groups. All tendons were explored 5 months after insertion. The vascular pedicle was patent in all vascularized tendons. Three tendon ruptures occurred in nonvascularized tendons and only one rupture occurred in a vascularized tendon. The digits with vascularized tendons demonstrated a significantly better simulated total active motion (117 degrees versus 128 degrees, p less than 0.05) than digits with nonvascularized tendons. The difference was even more significant when a localized tenolysis of the proximal juncture of the tendon graft was performed (215 degrees versus 138 degrees, p less than 0.01). This study supports the concept that vascularized tendon grafts may be advantageous in scarred tendon beds.