Bone regeneration and the osteoinductive capacity of implants are challenging issues in clinical medicine. Currently, recombinant growth factors and nonviral gene transfer are the most frequently investigated methods for bone growth enhancement, although the more favorable method remains unclear. There is a lack of knowledge in literature about the in vivo comparison of these methods for bone regeneration. BMP-2, which is the most commonly used growth factor for osteogenesis, was applied at its most efficient dose as a recombinant growth factor (rhBMP-2) and as a growth-factorencoding copolymer protected gene vector (pBMP-2) in a critical size bone defect (CSD) model to determine the most suitable method for bone regeneration. CSDs were induced bilaterally in 32 Sprague-Dawley rats. RhBMP-2 (62.5 μg) or pBMP-2 (2.5 μg) was embedded in poly(D,L-)lactide-coated titanium discs. Survival times were set at 14, 28, 56, and 112 days. After euthanasia, samples were analyzed via micro-computed tomography, polychrome sequential fluorescent labeling, and immunohistochemistry. Whereas defects in both groups were bridged with new bone after 56 days, rhBMP-2 initially induced ectopic new bone formation that was later remodeled in an unorganized hypodense manner. In contrast, pBMP-2 led to slower but steady bone regeneration with physiological tissue morphology, as confirmed by high osteoblast activity shown by osteocalcin staining. CD68 and TRAP staining verified high osteoclast activity for the rhBMP-2 group. pBMP-2 successfully induced locally controlled physiological bone regeneration, whereas rhBMP-2 triggered rapid and ectopic but insufficient bone formation. Thus, nonviral gene transfer appears to be more favorable for clinical applications.