Achieving high-efficiency tumor targeting after systemic delivery is a considerable challenge facing oncolytic gene therapists. Efficient retargeting should be combined with efforts to improve in vivo safety, reduce hepatotoxicity, minimize off-target interactions, and improve antitumoral potency and efficacy. We previously described the successful retargeting of adenovirus serotype 5 (Ad5) to α(v)β(6), an integrin that is highly overexpressed in numerous human carcinomas. In this study, we have further modified this construct by introducing mutations that ablate coxsackievirus-adenovirus receptor (CAR) binding and putative interactions with factor IX (FIX)/C4b-binding protein (C4BP). We have found that the resulting vector, Ad5-477dlTAYT(A20), displays a desirable in vivo safety profile. This vector does not agglutinate human erythrocytes, fails to cause thrombocytopenia after intravenous delivery, has limited induction of proinflammatory cytokines, and results in low-level toxicity (aspartate aminotransferase/alanine aminotransferase) when compared with Ad5-EGFP(WT). Furthermore, it has reduced accumulation in Kupffer cells (1 hr) and limited hepatocyte transduction at later time points (24 and 96 hr). The parental vector, Ad5-EGFP(A20), also displayed many of these desirable properties. As a result of the improved safety profile of both A20-modified vectors, we escalated the dose from 2×10(10) to 4×10(10) viral particles in an antitumoral efficacy study. We observed improvements in reducing percent tumor growth at early time points (96 hr) when compared with Ad5-EGFP(WT), although increasing the dose did not affect the therapeutic outcome beneficially. On completion of the experiment, we detected increased E1A staining in the tumors of all A20-treated groups and we determined that E1A expression was localized largely within α(v)β(6)(+) tumor cells. However, in spite of apparently efficient tumor transduction, this did not result in enhanced antitumoral efficacy as the virus failed to disseminate effectively throughout the tumor mass, presumably due to physical intratumoral restrictions. This highlights a remaining challenge that needs to be overcome before such vectors can be developed for future cancer gene therapy applications.