The human herpesviruses represent excellent candidate viruses for several types of gene vector applications. As a class, they are large DNA viruses with the potential to accommodate large or multiple transgene cassettes, and they have evolved to persist in a lifelong nonintegrated latent state without causing disease in the immune-competent host. Among the herpesviruses, herpes simplex virus type 1 (HSV-1) is an attractive vehicle because in natural infection, the virus establishes latency in neurons, a state in which viral genomes may persist for the life of the host as intranuclear episomal elements. The natural lifelong persistence of latent genomes in trigeminal ganglia (TG) without the development of sensory loss or histologic damage to the ganglion attests to the effectiveness of these natural latency mechanisms. Although the wild-type virus may be reactivated from latency under the influence of a variety of stresses, completely replication defective viruses can be constructed that retain the ability to establish persistent quiescent genomes in neurons, but that are unable to subsequently reactivate in the nervous system. These persistent genomes are devoid of lytic gene expression, but retain the ability to express latency-associated transcripts (LATs). We have made considerable progress in the construction of highly defective HSV genomic vectors that are apparently apathogenic for the nervous system, but are efficiently taken up by neurons to establish persistence (1 ).