Direct gene transfer to mammalian tissues has significant potential for biomedical research and gene therapy. Recently, the efficient transfer of naked plasmid DNA to the mouse liver by a rapid high-volume tail-vein injection was reported. We carried out a systematic analysis of the dose and time dependence of the expression of the Escherichia coli beta-galactosidase gene transferred by this technique. Surprisingly, the DNA concentration of the administered solution determined primarily the cellular gene dose and, hence, the expression of the transgene in individual hepatocytes, while the number of transfected cells was largely independent of the supplied plasmid mass. Transgene expression was transient: after a rapid onset and a peak at 8 h past injection, it gradually declined and was no longer detectable 4 weeks later. Although gene transfer was accompanied by tissue damage and subsequent regenerative proliferation, the decline in transgene expression was not due to increased hepatocyte turnover or to promoter downregulation, but instead cells apparently lost the plasmid DNA. Furthermore, we show that "nakedness" of the injected DNA is indeed a prerequisite for efficient transfer by the hydrodynamics-based procedure. Our data provide important clues for the successful use of this gene transfer technique, and may point directions for studies on the underlying mechanisms.