Anastomotic intimal hyperplasia and surface thrombogenicity are the main reasons for the high failure rate of prosthetic small‐diameter vascular grafts. While anastomotic intimal hyperplasia is a multifactorial event, ongoing surface thrombogenicity is primarily caused by the lack of an endothelium, even after years of clinical implantation. After decades of poorly performing synthetic artery‐grafts, tissue engineering has emerged as a promising approach to generate biologically functional bio‐synthetic hybrid grafts mimicking native arteries regarding the presence of an endothelial lining on the blood surface. “In vitro endothelialization” represented the first generation of such tissue‐engineered vascular grafts, utilising cell culture techniques for the creation of a confluent autologous endothelium on ePTFE grafts. The clinical long‐term results with this method in almost 200 patients are highly encouraging, showing patencies equal to vein grafts. Since “in vitro endothelialization” requires cell culture facilities, it will always be confined to large centres. Therefore, research of the 1990s turned to the development of spontaneously endothelializing implants, to make tissue‐engineered grafts amenable to the entire vascular‐surgical community. Apart from scaffold designs allowing transmural ingrowth, biological signalling through a facilitating ingrowth matrix holds a key to spontaneous endothelialization. In biological signalling, the increasingly deeper understanding of bio‐active molecules and the discovery of domains and peptide sequences during the 1980s created the expectation in the 1990s that peptide signalling may be all that is needed. This present review highlights the possible problems associated with such a reductionist approach. Using the fibronectin molecule, we demonstrated that domains may be more suitable modules in tissue engineering than peptide sequences. Anat Rec 263:379–387, 2001. © 2001 Wiley‐Liss, Inc.