Studies were conducted on smallholder plots in northern Togo from 1988 to 1991 to determine the possibility of applying treatments according to action thresholds. Thresholds were defined for the following pests: H. armigera, D. watersi, Earias sp., S. derogata and A. gossypii. The threshold programme was compared to the routine insecticide application programme involving five applications fortnightly beginning 50 d.a.s. An u.l.v. formulation of a mixture of deltamethrine and dimethoate at 3 litres/ha was applied, in both programmes. The results obtained showed that seed cotton yields were significantly different in 1991 (164 kg/ha less in the threshold treatment programme). Calculation shows that a reduction in treatments from five to three is not economical for the producer under the cropping conditions of the experiment. However, it could be worthwhile for the cotton-growing development company, if solutions are found to the numerous practical problems brought to light by the experiments conducted over the past 4 years.