CDC (Communicable Disease Center) light-traps were compared with human-bait collections as an alternative method for sampling malaria vectors in two villages of western Kenya. The numbers of Anopheles funestus Giles and Anopheles gambiae sensu lato Giles in CDC light-trap collections were significantly correlated to the numbers caught in human-bait collections, but in Anopheles arabiensis Patton the two collections were not significantly correlated. Most of the female vectors collected in the traps were unfed. Parity of A. arabiensis collected in CDC light-traps (44.3%) was significantly lower than the rate obtained from females caught on human-bait (54.5%). Although CDC light-traps provide a cheap and convenient method for collecting vectors, further studies should nevertheless be undertaken to determine the influence of species-specific variation in the sizes and age-structure of collections because such variations can affect the epidemiological interpretation of the data.