To improve the efficacy of a block copolymer of poly-d, l-lactic acid with randomly inserted p-dioxanone and polyethylene glycol (PLA-DX-PEG) as a drug delivery system for recombinant human bone morphogenetic proteins (rhBMPs), we examined the relationship between the volume of PLA-DX-PEG, the dose of rhBMP-2 and osteoinduction in a mouse model of ectopic bone formation. In a series of studies, we compared the size and bone mineral content (BMC) of ectopically induced bone by PLA-DX-PEG and collagen sponges carrying different quantities of rhBMP (0, 1, 2, 5, 10, 20 microg). An additional experiment was designed to investigate how a range of PLA-DX-PEG polymer volumes (15, 30, 60, 90 mg) with a fixed rhBMP concentration (0.01 wt%), altered the size and BMC of the induced ossicle. The influence of polymer volume was also examined in a further experiment wherein a fixed amount of rhBMP was placed in a range of PLA-DX-PEG copolymer volumes to give different concentrations of the protein per implant (0.02-0.0017 wt%). The results indicate that the bone yields were linearly dependent on the dose of rhBMP and also were proportional to the polymer volume above the minimal concentration of rhBMP-2 (0.0017 wt% in this series). The optimal concentration of rhBMP-2 in PLA-DX-PEG was 0.003 wt% in mice. The data provide important insights into the fabrication of implants that provide efficacious delivery of rhBMP-2 using the lowest possible dose of this expensive osteoinductive protein. This information will be of value for the clinical use of BMPs.