Host-seeking mosquitoes were sampled in bird-baited traps at four sites in New York state in 2003–2004. Trap placement and efficacy of chickens, Gallus gallus domesticus L., as bait compared with house sparrows, Passer domesticus L., an important reservoir of West Nile virus (family Flaviviridae, genus Flavivirus, WNV), was evaluated. Each site had a chicken-baited trap near ground level (≈1.5 m) and in the tree canopy (≈9 m), and a house sparrow-baited trap at ground level and canopy level. Each trap allowed mosquito access to birds on one end, and an inner mesh screen blocked bird access on the other end. The two most abundant mosquitoes, Culex restuans Theobald and Culex pipiens pipiens L., were differentiated using molecular characters. In 2003, Cx. restuans and Cx. p. pipiens made up 88% of total mosquito catch. In 2004, Cx. restuans comprised 43% of total catch and Cx. p. pipiens comprised 33%. The remaining species representing at least 1% of total catch were Ochlerotatus trivittatus (Coquillett), Coquilletidia perturbans (Walker), and Culiseta morsitans (Theobald). Capture rates were similar for chicken and house sparrow-baited traps; however, significantly more mosquitoes were captured in the canopy for both bird species. Cx. restuans preferred canopy traps, whereas equal numbers of Cx. p. pipiens were captured at ground and canopy levels. Mosquitoes were more likely to escape (74%) when excluded from birds than when allowed free access to birds (54%). Sentinel bird surveillance for WNV can be improved by trapping in the tree canopy in addition to ground level to capture the most important avian vectors.