Studies with first-generation adenoviral vectors have uncovered limitations that include finite transgene persistence, potential hepatotoxicity, and contamination with replication-competent adenovirus (RCA). To address these limitations within the context of cancer suicide gene therapy, a new adenoviral vector was developed containing the herpes simplex virus type 1 thymidine kinase (HSV tk) gene inserted in the E1 region of a recombinant vector containing deletions in the E1 and E4 regions of the Ad5 genome. The HSV tk minigene was placed under transcriptional control of a Rous sarcoma virus (RSV) promoter. This new E1E4-deleted vector was compared with the first-generation E1E3-deleted Ad.RSVtk vector. Generation of replication-competent adenovirus during production was eliminated. Using semiquantitative immunoblotting, the two vectors produced equivalent amounts of the expected 44-kDa tk-encoded protein in three different cell lines tested. The ability of the E1E4-deleted vector to sensitize tumor cells to ganciclovir (GCV) using in vitro assays and mixing studies was comparable to that of the E1E3-deleted vector. In vivo bystander effects were investigated using mixing studies in a syngeneic flank tumor model and demonstrated no difference between vectors in either immunocompetent or immunodeficient mice. To test the efficiency of these vectors in treating tumors in clinically relevant models, virus was injected intraperitoneally into tumor-bearing SCID mice and intrapleurally in a syngeneic rat mesothelioma model. After treatment of animals with ganciclovir, both vectors were roughly equivalent in their ability to increase mean survival (from approximately 40 to approximately 70 days) and markedly reduce tumor burden. Finally, formal toxicology studies were performed and showed similar amounts of local inflammation without systemic toxicity. In summary, this series of in vitro and in vivo experiments indicates that the performance of the recombinant E1E4-deleted adenoviral vector was virtually identical to that of the E1E3-deleted vector. Since the E1E4 vector has a much lower rate of recombination during production and has been shown to be less hepatotoxic in animal models, this new vector should prove superior to the first-generation Ad.HSVtk vectors in clinical cancer gene therapy trials.