Fibronectin is an extracellular matrix protein with broad binding specificity to cell surface receptors, integrins. The tenth fibronectin type III domain (FNfn10) is a small, autonomous domain of fibronectin containing the RGE sequence that is directly involved in integrin binding. However, in isolation FNfn10 only weakly bind to integrins. We reasoned that high-affinity and high-specificity variants of FNfn10 to a particular integrin could be engineered by optimizing residues surrounding the integrin-binding RGD sequence in the flexible FG loop. Affinity maturation of FNfn10 to alphavbeta3 integrin, an integrin up-regulated in angiogenic endothelial cells and in some metastatic tumor cells, yielded alphavbeta3-binding FNfn10 mutants with a novel RGDWXE consensus sequence. We characterized one of the RGDWXE-modified clones, FNfn10-3JCLI4, as purified protein. FNfn10-3JCLI4 binds with high affinity and specificity to purified alphavbeta3 integrin. Alanine scanning mutagenesis suggested that both the tryptophan and glutamic acid residues following the RGD sequence are required for maximal affinity and specificity for alphavbeta3. FNfn10-3JCLI4 specifically stained alphavbeta3-positive cells as detected with flow cytometry and it inhibited alphavbeta3-dependent cell adhesion. As with the anti-alphavbeta3 antibody LM609, FNfn10-3JCLI4 can interfere with in vitro capillary formation. Taken together, these data show that FNfn10-3JCL14 is a specific, high-affinity alphavbeta3-binding protein that can inhibit alphavbeta3-dependent cellular processes similar to an anti-alphavbeta3 monoclonal antibody. These properties, combined with the small, monomeric, cysteine-free and highly stable structure of FNfn10-3JCLI4, may make this protein useful in future applications involving detection and targeting of alphavbeta3-positive cells.