Although horse cases frequently are reported during West Nile virus (WNV) outbreaks, few investigations have focused on the epidemiology of this transmission. From April to October 2003 to 2005, mosquito abundance and infection were monitored 3 days per week at an equine research facility at the University of California, Davis. Thirty-two nonvaccinated horses enrolled as controls in a vaccine study were bled monthly, and their serum was tested for evidence of WNV infection by plaque reduction neutralization test (PRNT). In 2004, one positive Culex pipiens pool was associated with a single horse that presented with confirmed WNV disease in late September. The annual incidence of clinical and subclinical WNV infection in the nonvaccinated horses was 16%, with an apparent to inapparent ratio of 1:4 among infected horses. In 2005, two Culex tarsalis and two Cx. pipiens WNV-positive pools were associated with an equine infection incidence of 62%, with an apparent to inapparent ratio of 1:17. The majority (79%) of 70 blood-engorged Cx. pipiens fed on birds and the remaining on equines (21%). Conversely, Cx. tarsalis fed primarily on equines (n = 23, 74%), followed by birds (n = 7, 23%) and 1 (3%) fed on a lagomorph. These data indicated that nonvaccinated horses were a sensitive indicator of WNV activity and that their risk of infection was associated with the presence of infection in Cx. pipiens and Cx. tarsalis, which served as both enzootic and bridge vectors amplifying WNV among birds and transmitting WNV to horses.