This is a retrospective study of primary repairs of flexor pollicis longus in 16 children under 15 years of age. Patients with injuries to the median or ulnar nerve at the wrist, crush injuries, skin loss or fracture were excluded. Repairs were carried out within 24 hours using a modified Kessler technique. The mean follow-up was for two years. The final results were evaluated using the criteria of Buck-Gramko and Tubiana. They were good or excellent in all except one patient who had a secondary tendon rupture. When compared with the non-injured thumb, however, there was a significant decrease in active interphalangeal flexion (> 30°) in one-third of cases. A new method of assessment is proposed for the recovery of function of the flexor pollicis tendon which is more suitable for children. Postoperative immobilisation using a short splint had a negative effect on outcome. The zone of injury, an early mobilisation programme or concurrent injury to the digital nerve had no significant effect on the final result