Diphtheria toxin (DT) is a potent inhibitor of protein synthesis. As little asa single molecule of DT can result in cell-cycle independent cell death. This profound potency has led to difficulties in the development of DT as a suicide gene in cancer gene therapy, because toxicity appears to be related primarily to the fidelity of basal gene expression and the yield of viral titer. We evaluated the feasibility of prostate-specific DT gene therapy by cloning the catalytic domain (A chain) of DT under the control of the prostate-specific antigen (PSA) promoter, the PSA promoter and enhancer, or the cytomegalovirus promoter. The data on expression of DT from the plasmid constructs demonstrate that the basal level of DT gene expression determines the toxicity. To better test the potential therapeutic efficacy of DT suicide gene therapy, we first developed a DT-resistant adenoviral packaging line (293DTR). This allowed us to manufacture a relatively high titer adenoviral vector encoding the DT-A gene under the control of the PSA promoter and enhancer (Ad5PSE-DT-A) as well as an attenuated DT-A virus (Ad5PSE-tox176). In vitro studies showed that our viral constructs preferentially kill PSA-positive prostate cancer cells in the presence of exogenous androgen (R1881). In vivo studies showed that the nu/nu mice with PSA-positive cancer cell LNCaP xenograft treated with wild-type DT-A virus had a rapid regression of tumors and survived over a year without tumor progression, whereas the attenuated DT-A virus restricted tumor growth for only 1 month. The same constructs had no significant effect on the non-PSA-secreting cell line DU-145. These encouraging results suggest that DT-A viral gene transfer may ultimately have a therapeutic role in the treatment of advanced human prostate cancer.