This study evaluated 3D printed polycaprolactone (PCL) composite scaffold and recombinant human bone morphogenetic protein-2 (rhBMP-2), loaded either onto a PCL composite scaffold or implant surface, for vertical bone augmentation with implant placement. Three-dimensional printed PCL frames were filled with powdered PCL, hydroxyapatite, and β-tricalcium phosphate. RhBMP-2 was loaded to the PCL composite scaffolds and implant surfaces, and rhBMP-2 release was quantified for 21 days. Experimental implants were placed bilaterally on 20 rabbit calvaria, and the PCL composite scaffolds were vertically augmented. The randomly allocated experimental groups were divided by carrier and rhBMP-2 dosage as no rhBMP-2 (control), 5 μg rhBMP-2 loaded to PCL composite (Scaffold/rhBMP-2[5 μg]), 5 μg rhBMP-2 loaded to implant (Implant/rhBMP-2[5 μg]), 30 μg rhBMP-2 loaded to PCL composite (Scaffold/rhBMP-2[30 μg]), and 30 μg rhBMP-2 loaded to implant (Implant/rhBMP-2[30 μg]). Histologic and histometric analyses were conducted after 8 weeks. In both scaffold-loading and implant-loading, rhBMP-2 released initially rapidly, then slowly and constantly. Released rhBMP-2 totaled 23.02 ± 1.03% and 24.69 ± 1.14% in the scaffold-loaded and implant-loaded groups, respectively. There were no significant differences in histologic bone-implant contact (%). Peri-implant bone density (%) was significantly higher in the Scaffold/rhBMP-2(30 μg) and Implant/rhBMP-2(30 μg) groups. Total bone density (%) was not significantly different between the Scaffold/rhBMP-2(5 μg), Implant/rhBMP-2(5 μg), and control groups, or between the Scaffold/rhBMP-2(30 μg) and Implant/rhBMP-2(30 μg) groups, but was significantly higher in the Scaffold/rhBMP-2(30 μg) and Implant/rhBMP-2(30 μg) groups than in the controls. Three-dimensional printed PCL composite scaffold with rhBMP-2 produced vertical osteogenesis and osseointegration, regardless of rhBMP-2 loading to the PCL composite scaffold or implant surface.