Tumor cells genetically modified to express immunostimulatory molecules can produce high levels of antitumor immunity in rodent models. Although a number of clinical trials are currently in progress to assess the value of the approach in human disease, almost all require ex vivo transduction of cultured tumor cells with retroviral vectors. This process is not feasible for many human malignancies, hampering clinical evaluation of the approach. We have used an E1a,1b/E3 deletion mutant of adenovirus containing either the lacZ or the human interleukin-2 (IL-2) gene to transduce human neuroblastoma cells. This vector transduces fresh neuroblastoma cells and neuroblastoma cell lines with an efficiency of 80-90%, compared to an efficiency of 0-14% obtained with retroviral vectors. Cells transduced with the IL-2 adenovector produce up to 12,000 pg of IL-2/10(6) cells/24 hr. IL-2 adenovector-transduced neuroblasts are immunostimulatory; when they are cultured with patient lymphocytes, they increase the proportion of DR+ T cells and generate major histocompatibility complex (MHC) unrestricted cytotoxic effector cells active against parental (nontransduced) tumor cells. We conclude that IL-2 adenovector can be used to transduce freshly isolated human tumor cells efficiently, which will then produce immunomodulatory quantities of the cytokine. The use of adenoviral rather than retroviral vectors facilitates preparation of human tumor "vaccines" and these vectors are now being used in our clinical study of neuroblastoma patients.