We investigated the efficiency of bone morphogenetic protein (BMP)-2 released slowly from gelatin sponge for tracheal cartilage regeneration. A 1-cm gap was made in the mid-ventral portion of each of 10 consecutive tracheal cartilages. In the control group (n = 4), the resulting gap was left untreated. In the gelatin group (n = 4), plain gelatin was implanted in the gap. In the BMP-2 group (n = 4), gelatin containing 100 mu g BMP-2 was implanted. We euthanatized all dogs in each group at 1, 3, 6, and 12 months after the implantation, respectively, and then examined the implant site macro- and microscopically. in the BMP-2 group, regenerated fibrous cartilage and newly formed bone were observed at 1 and 12 months. Regenerated cartilage was observed at the ends of the host cartilage stumps, with newly formed bone in the middle portion. The gaps were filled with regenerated cartilage and newly formed bone. At 3 and 6 months, regenerated cartilage, but not newly formed bone, was evident. The regenerated cartilage was covered with perichondrium and showed continuity with the host cartilage. We succeeded in inducing cartilage regeneration and new bone formation in canine trachea by slow release of 100 mu g BMP-2 from gelatin.