Mosquitoes were caught in bedrooms in Tanzanian villages by human-biting catches and in light-traps set close to occupied untreated bed nets. Catches by each method were carried out on pairs of nights in the same week at different seasons and in different villages. The pairs of adjacent catches by the different methods showed a strong correlation. Analysis of the ratio between the catches by the two methods on pairs of nights in the same week indicated that on average three light-traps caught about the same number of mosquitoes as a team of two human catchers. The ratio did not differ significantly between Anopheles gambiae Giles (sensu lato), A. funestus Giles, and Culex quinquefasciatus Say, nor between the villages, or between times when mosquito populations were high or low. The distribution of numbers of ovarian dilatations differed significantly between catches in different villages and seasons but not between pairs of catches by the two methods. Similarly, the parity and sporozoite rates agreed between pairs of light-trap and house-resting catches, but differed markedly between villages and seasons. Thus it is concluded that light-traps used in conjunction with bed nets catch a representative sample of the vectors which would have bitten humans in bedrooms in this area.