Yeast cytosine deaminase (yCD)-based gene therapy offers the potential for selective production of the cytotoxic and radiosensitizing drug 5-fluorouracil (5-FU) from the benign prodrug 5-fluorocytosine within colorectal cancers. Although previous attempts to target therapy to colorectal cancer using the carcinoembryonic antigen (CEA) promoter have demonstrated specificity, this has been achieved at the cost of 10- to 300-fold loss in activity compared with strong but nonspecific rous sarcoma virus (RSV) or cytomegalovirus promoters. We developed a highly specific and active gene transfer method for colorectal cancer using CEA under control of a promoter-enhancer. We compared the RSV promoter-derived with the CEA promoter-enhancer-derived transgene expression in 10 different cell lines with differing CEA status. We found that the transgene expression resulting from both transient transfection and adenoviral infection with the CEA promoter-enhancer was as strong as the RSV promoter while maintaining specificity for CEA-producing cell lines. For instance, when we compared yCD expression between LoVo (CEA+) and human fibroblast (CEA-), we found a 30-fold-increased yCD expression in LoVo cells from CEA-enhancer adenovirus although there was no difference in the yCD expression between the cell lines when infected with RSV/yCD virus. This specificity was also achieved while maintaining a higher yCD enzyme activity than we obtained with RSV/yCD adenovirus in an HT-29 intrahepatic tumor model. We then compared the response of HT-29 xenografts to treatment with 5-fluorocytosine and yCD adenovirus driven by either the RSV or the CEA promoter-enhancer and found similar tumor growth inhibition. These findings suggest that the CEA promoter-enhancer strategy confers specificity while preserving activity and is worth exploring in additional animal and, potentially, clinical trials.