The intradermal administration of DNA vaccines by tattooing is a promising delivery technique for genetic immunization, with proven high immunogenicity in mice and in nonhuman primates. However, the parameters that result in optimal expression of DNA vaccines that are applied by this strategy to human skin are currently unknown. To address this issue we set up an ex vivo human skin model in which DNA vaccine-induced expression of reporter proteins could be monitored longitudinally. Using this model we demonstrate the following: First, the vast majority of cells that express DNA vaccine-encoded antigen in human skin are formed by epidermal keratinocytes, with only a small fraction (about 1%) of antigen-positive epidermal Langerhans cells. Second, using full randomization of DNA tattoo variables we show that an increase in DNA concentration,needle depth, and tattoo time all significantly increase antigen expression ( p < 0.001), with DNA concentration forming the most critical variable influencing the level of antigen expression. Finally, in spite of the marked immunogenicity of this vaccination method in animal models, transfection efficiency of the technique is shown to be extremely low, estimated at approximately 2 to 2000 out of 1 x 10(10) copies of plasmid applied. This finding, coupled with the observed dependency of antigen expression on DNA concentration, suggests that the development of strategies that can enhance in vivo transfection efficacy would be highly valuable. Collectively, this study shows that an ex vivo human skin model can be used to determine the factors that control vaccine-induced antigen expression and define the optimal parameters for the evaluation of DNA tattoo or other dermal delivery techniques in phase 1 clinical trials.