The invasion of different southern California landscapes by West Nile virus (WNV) and its subsequent amplification to epidemic levels during 2004 enabled us to study the impact of differing corvid populations in three biomes: the hot Colorado desert with few corvids (Coachella Valley), the southern San Joaquin Valley (Kern County) with large western scrub-jay but small American crow populations, and the cool maritime coast (Los Angeles) with a large clustered American crow population. Similar surveillance programs in all three areas monitored infection rates in mosquitoes, seroconversion rates in sentinel chickens, seroprevalence in wild birds, numbers of dead birds reported by the public, and the occurrence of human cases. Infection rates in Culex tarsalis Coquillett and sentinel chicken seroconversion rates were statistically similar among all three areas, indicating that highly competent mosquito hosts were capable of maintaining enzootic WNV transmission among less competent and widely distributed avian hosts, most likely house sparrows and house finches. In contrast, infection rates in Culex pipiens quinquefasciatus Say were statistically higher in Kern and Los Angeles counties with elevated corvid populations than in Coachella Valley with few corvids. Spatial analyses of dead corvids showed significant clusters near known American crow roosts in Los Angeles that were congruent with clusters of human cases. In this area, the incidence of human and Cx. p. quinquefasciatus infection was significantly greater within corvid clusters than without, indicating their importance in virus amplification and as a risk factor for human infection. In contrast the uniform dispersion by territorial western scrub-jays resulted in a high, but evenly distributed, incidence of human disease in Kern County.