In order to provoke an immune response, a tumor vaccine should not only maximize antigen-specific signals, but should also provide the necessary "co-stimulatory" environment. One approach is to genetically manipulate tumor cells to either secrete lymphokines (GM-CSF, IL-12, IL-15) or express membrane bound molecules (CD80, CD86). Furthermore, patient dendritic cells can be loaded with tumor-associated antigens or peptides derived from them and used for immunotherapy. Genetic modification of dendritic cells can also lead to presentation of tumor-associated antigens. Transfection of dendritic cells with DNA encoding for such antigens can be done in vitro, but transfection efficiency has been uniformly low. Alternatively, dendritic cells can also be modulated directly in vivo either by "naked" DNA immunization or by injecting replication-deficient viral vectors that carry the tumor specific DNA. Naked DNA immunization offers several potential advantages over viral mediated transduction. Among these are the inexpensive production and the inherent safety of plasmid vectors, as well as the lack of immune responses against the carrier. The use of viral vectors enhances the immunogenicity of the vaccine due to the adjuvant properties of some of the viral products. Recent studies have suggested that the best strategy for achieving an intense immune response may be priming with naked DNA followed by boosting with a viral vector. We have successfully completed a phase I and phase II clinical trials on immunotherapy of prostate cancer using naked DNA and adenoviral immunizations against the prostate-specific membrane antigen (PSMA) and phase I clinical trial on colorectal cancer using naked DNA immunization against the carcinoembryonic antigen (CEA). The vaccination was tolerated well and no side effects have been observed so far. The therapy has proven to be effective in a number of patients treated solely by immunizations. The success of the treatment clearly depends on the stage of the disease proving to be most efficient in patients with minimal disease or no metastases. A panel of changes in the phenotype of peripheral blood lymphocytes and the expression of intra-T-cell lymphokines seems to correlate with clinical improvement.