The effectiveness of DNA immunization has been demonstrated in several model systems, usually following intramuscular injection of DNA in saline, or topical administration to the skin. In this study we have compared DNA delivered by three routes (intramuscular, intravenous, and intraperitoneal) and, for each route, in two vehicles (cationic liposome complex and pH sensitive liposome). These two lipid vehicles were evaluated because they are frequently used in gene therapy studies, but their immunogenicity has not been extensively studied. Each of these six combinations has been evaluated not only by assay of marker gene expression in a variety of tissues, but also by measurement of biologically-relevant parameters of immunity induction of antibodies, cytotoxic T lymphocytes, and protection against viral challenge. By both criteria (marker gene expression and induced immunity), the outcomes vary markedly among the six combinations. The combination leading to maximal marker gene expression (DNA with cationic lipid, administered i.v.) also induces detectable antibodies and CTL, and is the only one of the six combinations to induce immune responses comparable to those seen following i.m. injection of DNA in saline. However, marker gene expression can be detected in other combinations in the absence of induced immunity thus the value of marker gene expression in predicting the protection induced by a microbial antigen is questionable suggesting that, when evaluating various promoter constructs, marker gene expression may not adequately replace the direct measurement of biological outcomes.