Highly pure, recombinant human osteoinductive proteins make it possible to consider programmable osteoneogenesis. Until recently, it was believed that a bioresorbable excipient or physiologic solution would suffice to transport osteoinductive agents from source to wound. After considering surgical requirements, particular bone wound circumstances, scarcity of collateral circulation, phenotype plasticity of mesenchymal progenitor cells, and the morphogens' pleiotrophic effects, it becomes clear that the issue of controlled, programmable osteoneogenesis is a more complicated proposition than can be addressed solely by application of osteoinductive protein. The essential characteristics of a manufactured bone graft substitute (BGS) device are dictated by demands placed on such a device by the surgeons who will employ them and the cells that will occupy them. This review outlines a design process for BGS devices that (1) begins by surveying BGS requirements gathered from the literature from 1991 to 1995, (2) briefly reviews recent in vitro studies of rhBMP-2 and OP- 1, (3) describes commonly encountered circumstances of recipient wound beds, (4) describes behaviors of mesenchymal cells involved in connective tissue repair and regeneration, and (5) concludes with a rationale for design of an osteoinductive bone graft substitute. Emerging from this process is a composite device consisting of a bioresorbable structural polymer, a filamentous velour of hyaluronan (HY), and an osteoinductive protein. The structural polymer, D,D-L,L-polylactic acid, fabricated in the architecture of cancellous bone, is capable of maintaining its structural and architectural properties after being thoroughly saturated with water. Within its interstices is located a filamentous velour of hyaluronan which, when fully hydrated, becomes a viscoelastic gel. It is anticipated that the osteoinductive protein will either be carried on the dried hyaluronic acid velour or in solution via the viscoelastic HY gel.