Treatment of injuries of the flexor digitorum profundus (FDP) tendon insertion site has changed little during the past 50 years, in part because there are no reports describing flexor tendon insertion site healing. Our objective was to assess the effects of repair technique and post-operative time on tendon–bone healing using a canine model of injury and repair. We transected 48 FDP tendons from 24 dogs at their insertions and repaired them using either a four- or eight-strand suture technique. We assessed the mechanical properties of the repaired tendon–bone construct, tendon collagen biochemistry, and distal phalanx bone mineral density (BMD) at 0, 10, 21 and 42 days. Suture method had no significant effect on any outcome (p > 0.05). In particular, use of an eight-strand double modified Kessler technique did not result in increased stiffness or strength compared to a four-strand technique. With time, the repair site became stiffer, as demonstrated by a 230% increase in rigidity and a 50% decrease in strain from 0 to 42 days. However, from 0 to 42 days the ultimate force of the insertion site did not increase. This lack of increase in ultimate force was consistent with decreases in collagen content, non-reducible crosslinks and distal phalanx BMD. Taken together, our results indicate that the canine FDP tendon heals slowly after it is injured at its insertion site and sutured onto the distal phalanx. While these findings may be limited to the particular repair method we used, they demonstrate a need for devising new treatment strategies to improve healing of flexor tendon insertion site injuries. © 2002 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.