Several approaches to the production of vaccines to human herpesviruses have been proposed. Subunit vaccines, subunits delivered by live vectors, and rationally attenuated vaccines have all been shown to be efficacious in animal models but suffer from uncertainties as to the roles of individual genes involved in pathogenesis and the most relevant components of the immune response required for protection in humans and the target antigens involved. With these problems in mind, we examined the vaccine potential of a fully disabled herpes simplex virus type 1 mutant that is capable of only a single round of replication, since a virus of this type should induce the full spectrum of immune responses but has no pathogenic potential. A virus has been described which lacks essential glycoprotein H (gH) and can be propagated in a cell line which supplies gH in trans (A. Forrester, H. Farrell, G. Wilkinson, J. Kaye, N. Davis-Poynter, and T. Minson, J. Virol. 66:341-348, 1992). Infection of normal cells with this mutant is indistinguishable from a wild-type infection, except that the resulting progeny are gH negative and noninfectious: the virus is self-limiting. Infection of mice by the ear pinna route was similarly self-limiting in that input infectivity decreased rapidly at the inoculation site and no infectivity was detected in sensory ganglia. Animals given a wide range of doses of the gH-negative mutant produced both humoral and T-cell responses to herpes simplex virus type 1 and proved solidly resistant to challenge with a high dose of wild-type virus. The gH-negative mutant is presumably capable of establishing a latent infection, but since no infectious virus was detected in numerous attempts to reactivate the mutant, the risk of a pathogenic outcome is minimal.