BACKGROUND The purpose of the study was to compare the efficacy on rat skull defects of two bone growth factors derived from the GDF-5 family. METHODS The study was conducted on 17 adult Wistar rats. On each animal, two symmetrical 6-mm wide, full-thickness, skull defects were carried out in the parietal regions. In 15 out of 17 animals, both experimental defects were filled by the implants. In the group I (n=2), both defects were left empty for control. The 15 other rats were divided into 3 groups: In group II (n=5), a collagen sponge was implanted. In group III (n=5), a collagen sponge impregnated with rhGDF-5 (the genuine dimeric form) was implanted. In group IV (n=5), a collagen sponge impregnated with rhGDF-5C465A (a monomeric form of GDF-5) was implanted. All animals were sacrificed at 8 weeks. The harvested specimens were processed for contact radiography and standard histological examination. The quantitative results were assessed with a semi-quantitative histological scoring system. RESULTS One animal in the group II was excluded because it died of unknown reasons. In group I, no bone healing was observed in the defects. In group II, no bone healing was observed in 4 out of 10 defects, and partial bone healing was observed in 5 out of 10 defects. In group III, partial bone healing was also observed in 3 out of 8 defects and complete bone healing in 4 out of 8 defects. In group IV, partial bone healing was observed in 8 out of 10 defects and complete bone healing in 2 out of 10 defects. CONCLUSIONS Bone healing was improved in all treated groups. Further studies are necessary to determine the optimal formulation of these composite implants.