The optimal release profile of locally delivered bone morphogenetic protein-2 (BMP-2) for safe and effective clinical application is unknown. In this work, the effect of differential BMP-2 release on bone formation was investigated using a novel biomaterial oligo[(polyethylene glycol) fumarate] bis[2-(methacryloyloxy) ethyl] phosphate hydrogel (OPF-BP) containing poly(lactic-co-glycolic acid) microspheres. Three composite implants with the same biomaterial chemistry and structure but different BMP-loading methods were created: BMP-2 encapsulated in microspheres (OPF-BP-Msp), BMP-2 encapsulated in microspheres and adsorbed on the phosphorylated hydrogel (OPF-BP-Cmb), and BMP-2 adsorbed on the phosphorylated hydrogel (OPF-BP-Ads). These composites were compared with the clinically used BMP-2 carrier, Infuse® absorbable collagen sponge (ACS). Differential release profiles of bioactive BMP-2 were achieved by these composites. In a rat subcutaneous implantation model, OPF-BP-Ads and ACS generated a large BMP-2 burst release (>75%), whereas a more sustained release was seen for OPF-BP-Msp and OPF-BP-Cmb (~25% and 50% burst, respectively). OPF-BP-Ads generated significantly more bone than did all other composites, and the bone formation was 12-fold higher than that of the clinically used ACS. Overall, this study clearly shows that BMP-2 burst release generates more subcutaneous bone than do sustained release in OPF-BP-microsphere composites. Furthermore, composites should not only function as a delivery vehicle but also provide a proper framework to achieve appropriate bone formation.