Established treatments such as surgery, radiation, and chemotherapy have only minimally altered the median survival time of patients with glioblastoma multiforme, the most common malignant brain tumor. These failures reflect the highly invasive nature of the disease, as well as the fact that few cells are actively dividing at any given time. As a result, therapies need to act in areas of the brain that are spatially separated from the site of tumor origin and over extended periods of time temporally separated from their introduction. Over the past decade, laboratory studies and early clinical trials have raised the hope that these therapeutic requirements may be fulfilled by gene therapy in which nonreplicating transgene-bearing viruses, oncolytic viruses, or migratory stem cells are used to deliver tumoricidal transgenes. The authors review the principles behind these approaches and their initial results.