Bone morphogenetic proteins (BMP), in particular BMP-2, are the growth factors primarily responsible for osteoinduction. A knowledge of interactions between bone substitute materials and growth factor variants is crucial to designing bone substitutes with an ideal release profile. Here we compare glycosylated and non-glycosylated recombinant human bone morphogenetic protein-2 (rhBMP-2) either incorporated into a hydrolyzable polyethylene glycol (PEG) hydrogel developed as a slow release system or adsorbed to a deproteinized bovine bone matrix (DBBM), a clinically well-established bone substitute material. rhBMP-2 loaded materials were immersed in cell culture medium and rhBMP-2 concentration profiles in the supernatant were determined by an enzyme-linked immunosorbent assay. The corresponding biological activities were assessed in vitro by alkaline phosphatase activity assay. We show a strong affinity of rhBMP-2 for DBBM and reduced biological activity after its release from PEG hydrogels. Glycosylated rhBMP-2 was significantly less affected by the hydrogel and interacted significantly more strongly with DBBM than non-glycosylated rhBMP-2. We therefore question the combination of PEG hydrogels with DBBM as a rhBMP-2 delivery system over DBBM alone, since rhBMP-2 released from the hydrogel will be trapped by DBBM. Moreover, our results suggest that glycosylated rhBMP-2 is favorable in combination with PEG hydrogels, since its activity is better preserved, whereas in combination with DBBM non-glycosylated rhBMP-2 is favorable, benefiting from an initially higher concentration of free rhBMP-2.