West Nile virus surveillance was conducted at five sites in New Castle County, DE, and one site in Salem County, NJ, from June through September, 2004, using dry ice-baited Centers for Disease Control miniature light traps, infusion-baited gravid traps, and resting boxes. All trap types were simultaneously placed at each site every two weeks and run overnight. Collected mosquitoes were identified to species, pooled, and analyzed for virus using a real-time reverse transcriptase polymerase chain reaction test. In total, 47, 972 mosquitoes in 29 species or species groups were analyzed. Light traps collected 60, 201 mosquitoes in 28 species or species groups. Gravid traps collected 3, 195 mosquitoes in 19 species or species groups. Resting boxes collected 99 mosquitoes in nine species or species groups. In total, 1, 500 mosquito pools were tested for WNV resulting in ten positive pools. All positive pools consisted of Culex pipiens, Culex restuans, or Culex salinarius. Seven positive pools were from gravid traps and three were from light traps despite testing almost 14 times as many pools from light traps. The overall infection rate from gravid traps was nearly 33 times greater than the infection rate from light traps, 2.29 and 0.07 infected mosquitoes per 1, 000, respectively. The results demonstrate the advantage of using gravid traps for West Nile virus surveillance over light traps or resting boxes.