Abstract. Laboratory tests of insect repellents by various different methods showed that An.stephensi Liston was consistently more susceptible than An.gambiae Giles, An.albimanus Wiedemann or An.pulcher-rimus Theobald. The six repellents tested were di-ethyl toluamide (deet), di-methyl phthalate (DMP), ethyl-hexanediol, permethrin, citronella and cedarwood oil. Testing systems in which the mosquitoes were presented with a choice gave consistently lower ED50 values than when there was no choice, i.e. the standards of tolerance are not absolute but depend on the options available. In field tests in an experimental hut a curtain with a high dose of di-ethyl toluamide (deet) reduced biting in the hut but had to be re-impregnated frequently. Deet-impregnated anklets gave about 84% protection against Culex quinquefasciatus Say for 80 days after one impregnation, in a trial in which the anklets were brought out of sealed storage and tested for 2 h nightly. Similar protection was found against An.funestus Giles but the protection against An.gambiae s.L, An.coustani Laveran and Mansonia spp. was not as good. There were highly significant differences between the four collectors' mosquito attractiveness but this varied highly significantly between the mosquito species.