Oncolytic virotherapy exploits the ability of viruses to infect, replicate into, and kill tumor cells. Among the viruses that entered clinical trials are HSVs. HSVs can be engineered to become tumor-specific by deletion of selected genes or retargeting to tumor-specific receptors. A clinically relevant surface molecule is HER-2, hyperexpressed in one fourth of mammary and ovary carcinomas, and associated with high metastatic ability. As a previously undescribed strategy to generate HSV recombinants retargeted to HER-2 and detargeted from natural receptors, we replaced the Ig-folded core in the receptor-binding virion glycoprotein gD with anti-HER-2 single-chain antibody. The recombinant entered cells solely via HER-2 and lysed HER-2–positive cancer cells. Because of the high specificity, its safety profile in i.p. injected mice was very high, with a LD ₅₀ >5 × 10 ⁸ pfu, a figure at least 10,000-fold higher than that of corresponding WT-gD carrying virus (LD ₅₀ ≈ 5 × 10 ⁴ pfu). When administered intratumorally to nude mice bearing HER-2–hyperexpressing human tumors, it strongly inhibited progressive tumor growth. The results provide a generally applicable strategy to engineer HSV recombinants retargeted to a wide range of receptors for which a single-chain antibody is available, and show the potential for retargeted HSV to exert target-specific inhibition of human tumor growth. Therapy with HER-2-retargeted oncolytic HSV could be effective in combined or sequential protocols with monoclonal antibodies and small inhibitors, particularly in patients resistant to HER-2-targeted therapy because of alterations in HER-2 signaling pathway, or against brain metastases inaccessible to anti-HER-2 antibodies.