Type I interferons (alpha/beta) have significant antitumor activity although their short half-life and systemic side effects have limited their clinical utility. An alternative dosing schedule of continuous, low-level delivery, as is achieved by gene therapy, rather than intermittent, high concentration pulsed-dosing, might avoid the toxicity of interferon while maintaining its antitumor efficacy. We have tested a gene therapy approach in murine tumor models to treat malignancies that have shown responsiveness to interferon in clinical trials. The tumor cell lines used were moderately sensitive to the direct effects of human interferon-beta (hIFN-beta) in vitro. For in vivo testing, systemic delivery of hIFN-beta was generated following liver-targeted delivery of adeno-associated virus (AAV) vector carrying the hIFN-beta transgene. This prevented engraftment of subcutaneous human gliomas, and orthotopic, localized (intrarenal) and disseminated (primarily pulmonary) human renal cell carcinomas; and caused regression of established tumors at these sites. In a syngeneic, immunocompetent model of melanoma, AAV IFN-beta treatment limited subcutaneous tumor growth and prevented disseminated disease. A significant decrease in mean intratumoral vessel density was demonstrated in hIFN-beta-treated tumors, suggesting that in addition to a direct tumoricidal effect, the antitumor efficacy of AAV IFN-beta in this study was due to its ability to inhibit angiogenesis.