Oncolytic replicating adenoviruses are a promising new modality for the treatment of cancer. Despite the assumed biologic advantage of continued viral replication and spread from infected to uninfected cancer cells, early clinical trials demonstrate that the efficacy of current vectors is limited. In xenograft tumor models using immune-incompetent mice, wild-type adenovirus is also rarely able to eradicate established tumors. This suggests that innate immune mechanisms may clear the virus or that barriers within the tumor prevent viral spread. The aim of this study was to evaluate the kinetics of viral distribution and spread after intratumoral injection of virus in a human tumor xenograft model. After intratumoral injection of wild-type virus, high levels of titratable virus persisted within the xenograft tumors for at least 8 weeks. Virus distribution within the tumors as determined by immunohistochemistry was patchy, and virus-infected cells appeared to be flanked by tumor necrosis and connective tissue. The close proximity of virus-infected cells to the tumor-supporting structure, which is of murine origin, was clearly demonstrated using a DNA probe that specifically hybridizes to the B1 murine DNA repeat. Importantly, although virus was cleared from the circulation 6 hr after intratumoral injection, after 4 weeks systemic spread of virus was detected. In addition, vessels of infected tumors were surrounded by necrosis and an advancing rim of virus-infected tumor cells, suggesting reinfection of the xenograft tumor through the vasculature. These data suggest that human adenoviral spread within tumor xenografts is impaired by murine tumor-supporting structures. In addition, there is evidence for continued viral replication within the tumor, with subsequent systemic dissemination and reinfection of tumors via the tumor vasculature. Despite the limitations of immune-incompetent models, an understanding of the interactions between the virus and the tumor-bearing host is important in the design of effective therapies.