Previous studies have demonstrated the ability of direct adenoviral BMP-2 (Ad.BMP-2) gene delivery to enhance bone repair. Nevertheless, in studies using a rat segmental defect model, it has not proved possible to achieve reliably full osseous union in all animals. To address this issue, we evaluated the effect of vector dose on healing. Critical-size defects were created in the right femora of 27 Sprague-Dawley rats. The defects received a single, intralesional, percutaneous injection of 2.7 x 10(7) (low dose), 2.7 x 10(8) (medium dose), or 2.7 x 10(9) (high dose) plaque-forming units of Ad.BMP-2. After 8 weeks, femora were evaluated by X-ray, dual-energy X-ray absorptiometry, microcomputed tomography (microCT), and histology. The high dose of vector bridged 100%, the medium dose 11%, and the low dose 25% of the defects, as evaluated by X-ray and microCT imaging. Bone mineral content and bone volume of the defects receiving the high dose of vector were significantly higher than those of both groups receiving lower doses. Histologically, defects treated with the high dose were filled by trabecular bone and small amounts of cartilage, whereas large areas of fibrous tissue and cartilage remained in the defects receiving lower doses. However, the newly formed bone lacked the structural organization of native bone, suggesting that further maturation is necessary.