With an increasing incidence of prostate cancer, identification of new tumor drivers or modulators is crucial. Genetically engineered mouse models (GEMM) for prostate cancer are hampered by tumor heterogeneity and its complex microevolution dynamics. Traditional prostate cancer mouse models include, amongst others, germline and conditional knockouts, transgenic expression of oncogenes, and xenograft models. Generation of de novo mutations in these models is complex, time-consuming, and costly. In addition, most of traditional models target the majority of the prostate epithelium, whereas human prostate cancer is well known to evolve as an isolated event in only a small subset of cells. Valuable models need to simulate not only prostate cancer initiation, but also progression to advanced disease. Here we describe a method to target a few cells in the prostate epithelium by transducing cells by viral particles. The delivery of an engineered virus to the murine prostate allows alteration of gene expression in the prostate epithelia. Virus type and quantity will hereby define the number of targeted cells for gene alteration by transducing a few cells for cancer initiation and many cells for gene therapy. Through surgery-based injection in the anterior lobe, distal from the urinary track, the tumor in this model can expand without impairing the urinary function of the animal. Furthermore, by targeting only a subset of prostate epithelial cells the technique enables clonal expansion of the tumor, and therefore mimics human tumor initiation, progression, as well as invasion through the basal membrane. This novel technique provides a powerful prostate cancer model with improved physiological relevance. Animal suffering is limited, and since no additional breeding is required, overall animal count is reduced. At the same time, analysis of new candidate genes and pathways is accelerated, which in turn is more cost efficient.