The dynamics of West Nile virus (family Flaviviridae, genus Flavivirus, WNV) infection in mosquitoes, sentinel pheasants, and wild dead birds were evaluated during 2003-2006 in Saginaw Co., MI. Mosquitoes were collected by New Jersey Light Traps at 22 sites during May-September, pooled by species and sample location, and tested for presence of WNV RNA by using a real-time reverse transcription-polymerase chain reaction assay. Oral swabs from wild dead birds submitted by the public were tested by Vec-Test assay. Sentinel pheasants were bled weekly, and serum was tested for antibodies with an inhibition enzyme immunoassay. In total, 37,225 mosquitoes [Aedes vexans (Meigen), Culex pipiens L., and Culex restuans Theobald] were tested in 5,429 pools, of which 59 (1.1%) were positive. Ae. vexans was most abundant but had a comparatively low infection rate (0.06-2.11) compared with Cx. pipiens (1.75-4.59) and Cx. restuans (1.22-15.67). Mosquito abundances were temporally related to variations in 2-wk average weather variables. Infected dead crows appeared earlier each transmission season than blue jays, but infection prevalence for both peaked approximately mid-August. Space-time clusters were found in different locations each year. Sentinel pheasant seroprevalence was 19.3% (16/83), 12.7% (10/79), and 7.7% (5/65) during 2003-2005, respectively. We demonstrated temporal patterns of WNV activity in corvid birds and Culex spp. mosquitoes during the study period, suggesting virus transmission within an enzootic cycle. Despite the absence of human case reports nearby, this surveillance system demonstrated WNV transmission and possible human risk. Maintained surveillance using more appropriate gravid traps and CDC CO2 light traps could improve sensitivity of vector collection and virus detection.