MALIGNANT GLIOMAS ARE the most common malignant brain tumors and are almost universally fatal. A genetically engineered herpes simplex virus-1 mutant with decreased neurovirulence, dlsptk, has been shown to kill human glioma cells in culture and in animal models. However, intracranial inoculation of dlsptk is limited by fatal encephalitis at higher doses. Therefore, additional engineered and recombinant herpes simplex mutants with demonstrated reduced neurovirulence (AraAr9, AraAr13, RE6, R3616) were examined as antiglioma agents. One long-term human glioma cell line and two early-passage human gliomas in culture were destroyed by all four viruses tested. In a subcutaneous glioma model, AraAr13, RE6, and R3616 retained substantial antineoplastic effects in nude mice when compared with controls (one-sided Wilcoxon rank test, P < 0.05 for one or more doses each). When tested in a nude mouse intracranial glioma model, both RE6 and R3616 significantly prolonged average survival without producing premature encephalitic deaths at two doses (log-rank statistic, P < 0.007). Histopathological studies of the brains of surviving animals revealed minimal focal encephalitis in two of three RE6-treated animals and no evidence of encephalitis in either one of three RE6-treated or in three of three R3616-treated animals. No evidence of residual tumor was seen in four of the six surviving animals. Additionally, both RE6 and R3616 were found to be susceptible to the common antiherpetic agent acyclovir, adding to their safety as potential antiglioma agents. Recombinant and engineered viruses that minimize host toxicity and maximize tumoricidal activity merit further study as antineoplastic agents. © by the Congress of Neurological Surgeons.