The response of Anopheles gambiae complex mosquitoes to men sleeping under insecticide-impregnated or untreated bednets in six verandah trap huts was studied during the dry season in The Gambia. With this type of hut it was possible to collect live and dead indoor-resting mosquitoes and estimate the number of wild mosquitoes which entered, bloodfed on man, and exited each night. Bednets were treated with emulsions targetted to leave deposits of 25 mg/m2 lambda-cyhalothrin, or 5, 50 or 500 mg/m2 permethrin, diluted from emulsifiable concentrates (EC), or a blank formulation similar to the EC except that the permethrin was omitted; the sixth net was left untreated. Nets and sleepers were rotated between huts on different nights, the design being based on a series of Latin squares and conducted double-blind. Permethrin-impregnated bednets deterred mosquitoes from entering the huts. The degree of deterrency was proportional to the dosage of permethrin. This effect was also caused by the blank formulation and therefore attributed to other components of the formulation, rather than to the permethrin itself. The net impregnated with 500 mg permethrin per square metre gave the best individual protection, reducing mosquito bloodfeeding by 91% compared with untreated nets. However, lambda-cyhalothrin was proportionately more insecticidal than permethrin at doses of equivalent deterrency. At this stage of research, it remains conjectural whether chemical deterrency or killing of malaria vectors is better for community protection.