Stromal-epithelial interaction contributes to local prostate tumor growth, androgen-independent progression and distant metastasis. We have established in vitro coculture and in vivo chimeric tumor models to evaluate the roles of stromal cells isolated from either osteosarcoma or normal bone, a site where prostate cancer cells frequently metastasize, in contributing to the growth and survival of human prostate cancer cells. We have evaluated extensively the effects of toxic gene therapy using luciferase-tagged chimeric human prostate cancer models both in vitro and in vivo. In the in vitro cocultured cell model, we assessed cancer cell growth and residual cellular proteins after targeting either prostate cancer epithelial cells alone or both prostate cancer and bone stromal cells. In the in vivo animal model, we measured tumor volume and serum prostate-specific antigen (PSA) in mice bearing chimeric prostate tumors comprised of human prostate tumor cells and normal bone stromal cells. Our results demonstrated that: (1) The rate of human prostate cancer cell growth in vitro is accelerated by coculturing with human and rat osteosarcoma or normal mouse bone marrow stromal cell lines. No growth stimulation was noted when cocultured with a human prostate epithelial cell line. (2) Disabling the growth of normal bone stromal cells using transgenic targeting with a bystander gene, herpes simplex virus thymidine kinase (hsv-TK), plus the pro-drug ganciclovir (GCV) or acyclovir markedly depressed the growth of cocultured human prostate cancer cells in vitro and human prostate cancer-mouse normal bone stroma chimeric tumors in vivo. (3) By cotargeting both human prostate cancer and normal mouse bone stromal cells in vitro with an adenoviral construct, Ad-hOC-TK (a replication-defective Ad5 vector with the bystander transgene hsv-TK under the control of a human osteocalcin (hOC) promoter) plus GCV4, we observed greater inhibition of tumor cell growth than by targeting a single cell compartment with Ad-PSA-TK (a vector construct similar to Ad-hOC-TK except that the transgene expression is under regulation by a full-length human PSA promoter). These results, taken together, established a basic principle that cotargeting both tumor and its supporting stroma is more efficacious than targeting a single cell compartment in the treatment of human prostate cancer bone metastasis. This principle can be applied to other clinical conditions of cancer growth where stroma contribute to the overall growth and survival potential of the cancer.