A growing number of antiangiogenesis strategies have been investigated for the treatment of cancer and other angiogenesis-dependent diseases. One of the most promising strategies is to systemically administer one or more antiangiogenic proteins frequently enough to achieve a sufficient long-term steady state level of the protein(s) to achieve the maximum beneficial effect. However, the utility of this strategy is limited because of many technical difficulties, including obtaining both the quantity and quality of the protein(s) necessary for optimal therapeutic benefit. To overcome these difficulties, we hypothesized that a single administration of a replication-defective adenoviral vector expressing a secretable antiangiogenic protein could achieve an optimal long-term systemic concentration. We constructed a recombinant adenoviral vector, Av3mEndo, which encodes a secretable form of murine endostatin. We demonstrated secretion of endostatin from several cell lines transduced with Av3mEndo. Partially purified endostatin secreted from Av3mEndo-transduced mammalian cells was shown to potently inhibit endothelial cell migration in vitro. A single intravenous administration of Av3mEndo in mice was shown to result in (1) prolonged and elevated levels of circulating endostatin, (2) partial inhibition of VEGF-induced angiogenesis in a VEGF implant angiogenesis model, and (3) prolonged survival and in 25% of mice the complete prevention of tumor growth in a prophylactic human colon/liver metastasis xenograft murine model. These results support our contention that adenoviral vector-mediated expression of an antiangiogenic protein(s) represents an attractive therapeutic approach to cancer and other angiogenesis-dependent diseases.