The vertical distribution of blood-feeding flies in two temperate forests in the southeastern U.S.A. was determined by placing 15 Centers for Disease Control and Prevention miniature light traps (12 CO(2)-baited, three unbaited controls), without lights, at three heights (1.5 m, 5.0 m, 10.0 m). More than 6550 haematophagous flies, representing 49 species in four families, were collected. Eighteen species were taken almost exclusively (90-100%) at 1.5 m or 10.0 m, and the mean number of flies per trap differed significantly with height for another six species. Five species exhibited shifts in vertical distribution between the two forests, indicating that forest structure could influence the height of host searching. Most (52.5%) mammalophilic flies were collected at 1.5 m, whereas most (56.4%) ornithophilic flies were taken at 10.0 m, suggesting that host associations influence vertical distributions. The significant differences in the composition of haematophagous fly populations among forest strata emphasize the importance of trap placement in vector surveillance and of understanding the ecological relationships of blood-feeding flies.