The goal of this work was to develop a growth factor delivery system for use in wound healing that would provide localized release of heparin-binding growth factors in a biomimetic manner, such that release occurs primarily in response to cell-associated enzymatic activity during healing. A key element of the drug delivery system was a bi-domain peptide with an N-terminal transglutaminase substrate and a C-terminal heparin-binding domain, based on antithrombin III. The bi-domain peptide was covalently cross-linked to fibrin matrices during coagulation by the transglutaminase activity of factor XIIIa and served to immobilize heparin electrostatically to the matrix, which in turn immobilized the heparin-binding growth factor and slowed its passive release from the matrix. Basic fibroblast growth factor (bFGF) was considered as an example of a heparin-binding growth factor, and cell culture experimentation was performed in the context of peripheral nerve regeneration. A mathematical model was developed to determine the conditions where passive release of bFGF would be slow, such that active release could dominate. These conditions were tested in an assay of neurite extension from dorsal root ganglia to determine the ability of the delivery system to release bioactive growth factor in response to cell-mediated processes. The results demonstrated that bFGF, immobilized within fibrin containing a 500-fold molar excess of immobilized heparin relative to bFGF, enhanced neurite extension by up to about 100% relative to unmodified fibrin. A variety of control experiments demonstrate that all components of the release system are necessary and that the bi-domain peptide must be covalently bound to the fibrin matrix. The results thus suggest that these matrices could serve as therapeutic materials to enhance peripheral nerve regeneration through nerve guide tubes and may have more general usefulness in tissue engineering.