Several polycations possessing substantial buffering capacity below physiological pH, such as lipopolyamines and polyamidoamine polymers, are efficient transfection agents per se--i.e., without the addition of cell targeting or membrane-disruption agents. This observation led us to test the cationic polymer polyethylenimine (PEI) for its gene-delivery potential. Indeed, every third atom of PEI is a protonable amino nitrogen atom, which makes the polymeric network an effective "proton sponge" at virtually any pH. Luciferase reporter gene transfer with this polycation into a variety of cell lines and primary cells gave results comparable to, or even better than, lipopolyamines. Cytotoxicity was low and seen only at concentrations well above those required for optimal transfection. Delivery of oligonucleotides into embryonic neurons was followed by using a fluorescent probe. Virtually all neurons showed nuclear labeling, with no toxic effects. The optimal PEI cation/anion balance for in vitro transfection is only slightly on the cationic side, which is advantageous for in vivo delivery. Indeed, intracerebral luciferase gene transfer into newborn mice gave results comparable (for a given amount of DNA) to the in vitro transfection of primary rat brain endothelial cells or chicken embryonic neurons. Together, these properties make PEI a promising vector for gene therapy and an outstanding core for the design of more sophisticated devices. Our hypothesis is that its efficiency relies on extensive lysosome buffering that protects DNA from nuclease degradation, and consequent lysosomal swelling and rupture that provide an escape mechanism for the PEI/DNA particles.