Insecticidal smoke generators were tested in the laboratory and in practice. Each generator contained 110–120 g. of crude insecticide, either D.D.T. or benzene hexachloride. Over half the insecticide was evolved unchanged as a smoke of particles less than 1 μ in diameter, by a slow combustion mixture. In the laboratory tests, the generators were used at the rate of one to 1800–2500 cu.ft. (50–70 cu.m.). Benzene hexachloride had a greater initial insecticidal effect and a better residual action up to a week after treatment; thereafter (1 and 6 months) the D.D.T. was more effective. The residual effects of horizontal surfaces were good against bedbugs, but the deposits on vertical surfaces were not highly efficient. Residual action against mosquitoes of vertical surfaces a week after treatment was fair, but effects of inverted surfaces were poor.The deposits of active insecticide on horizontal surfaces were about 15–30 mg./sq.ft.; on vertical surfaces, 3–10 mg./sq.ft.; and on inverted surfaces, 2–6 mg./sq.ft.A field trial was conducted between August and November in kitchens infested by flies and cockroaches. The generators were used at the rate of one per 64 cu.ft. (18 cu.m.). A substantial reduction of flies for 11 weeks was obtained by D.D.T., but for only 2 weeks with benzene hexachloride. Both treatments reduced heavy cockroach infestations to negligible proportions for 2–3 months. There was evidence of some repellent effect.Thermal generators require careful precautions to prevent indirect tainting of food or beverages. They are very wasteful of insecticide especially since two or perhaps three treatments per season might be necessary. This would use twenty times as much insecticide as a spray treatment. On the other hand, labour and equipment costs are somewhat lower, and ease of operation very much greater, with sm