Herpes simplex virus 1 has a number of biological features which suggest that it could be engineered as a vector for direct transfer of therapeutic genes to neurons. These features include (i) its natural ability to establish life-long latency, a state in which the viral genome is not integrated, lytic genes are quiescent, and the metabolic functioning of the host cell is apparently undisturbed; (ii) the expression of latencyassociated transcripts (LATs) driven by neuron-specific, latency-active promoter (LAP) elements, which may prove useful in expressing transgenes from latent viral genomes; and (iii) the observation that replication-defective mutants created by the deletion of essential genes retain the ability to establish a latent state in the nervous system (1). In addition, many of the 81 herpes simplex virus (HSV) genes are not required for viral replication in cell culture and may conveniently be deleted to provide space for incorporation of substantial foreign DNA, and almost all viral genes are contiguous units, making genetic manipulation feasible. The virus can also be grown to high titer, and viral infectivity is very efficient. The major impediments to the development of HSV-effective vectors relate to residual cytotoxicity of defective vectors and the limited duration of transgene expression. Even replication-incompetent mutant viruses are cytotoxic, readily killing neurons in vitro, and with the exception of the HSV LAP elements, viral and foreign promoters appear to come under control of the virus' ability to rapidly induce mechanisms of promoter shutoff. Two different types of HSV-based gene delivery systems have been developed. The type consists of genetically engineered genomic vectors, which may be deleted in genes required for the virus to replicate in postmitotic cells such as neurons or may be completely replication-defective, requiring complementation for vector propagation. The second type of HSV-based vector system, referred to as amplicons, uses defective helper-virus mutants for packaging concatemeric plasmids containing an HSV origin of DNA synthesis and a packaging sequence. We have focused our efforts on the development of replication-defective genomic vectors. The first generation defective genomic vectors were deleted in the single essential immediate early (IE) gene encoding ICP4 (e.g., d120) (2). These vectors can be propagated in ICP4-complementing cell lines, but on infection of neurons, viral gene expression is aborted at the level of IE gene expression. Although these vectors are of reduced pathogenicity and can be used to efficiently transfer and transiently express reporter genes in brain (see below), they are toxic to neurons in culture, producing cytopathic effects such as cytoplasmic blebbing, host cell DNA fragmentation, and chromo-