Epidermal growth factor receptor (EGF-R) is an important target in anticancer therapy. Here we report how a novel EGF-R peptide ligand (D4: Leu-Ala-Arg-Leu-Leu-Thr) is identified using a computer-aided design approach from a virtual peptide library of putative EGF-R binding peptides by screening against the EGF-R X-ray crystal structure in silico and in vitro. The selected peptide is conjugated with a polyethylene glycol (PEG) lipid, and the lipid moiety of the peptide-PEG-lipid conjugate is inserted into liposome membranes by a postmodification process. D4 peptide-conjugated liposomes are found to bind to and enter cells by endocytosis specifically and efficiently in vitro in a process apparently mediated by EGF-R high-expressing cancer cells (H1299). In vivo, the D4 peptide-conjugated liposomes are found to accumulate in EGF-R-expressing xenograft tumor tissues up to 80 h after intravenous delivery, in marked contrast to controls. These results demonstrate how structure-based peptide design can be an efficient approach to identify highly novel binding ligands against important receptors. These data could have important consequences for the development of peptide-directed drug delivery systems with engineered specificities and prolonged times of action.