Insecticides were tested in the glasshouse and in small field plots. Aldicarb, chlorpyrifos, DDT and HCH, applied to the soil surface in pots decreased midge emergence from larval cocoons. When infested canes were sprayed in the laboratory, chlorpyrifos and fenitrothion killed more larvae than DDT, HCH or permethrin. In the first field trial, two sprays of fenitrothion in August to the basal 60 cm of canes damaged by a machine harvester decresed the numbers of larvae per cane by 93%. In subsequent trials insecticides were applied to primocanes, both in conventionally managed rows and in rows managed in a biennial cropping system. Two sprays of chlorpyrifos, diazinon or fenitrothion applied against first generation larvae in May and June decreased the subsequent population of overwintering larvae. In the following year these insecticides, and also HCH and triazophos, all applied three times in May and June controlled first generation larvae. However, lesion development, caused by fungi penetrating tissue beneath feeding sites of second generation larvae in July and August, was not decreased by these treatments and they did not increase yield, probably due to adult midges from unsprayed rows re-infesting treated plots. Two applications of HCH in May and June increased yield the following year. In the biennially cropped plots midge and midge blight were controlled and yields were increased by insecticides in the first year when blight killed 63% of untreated canes, but infestations were too low to demonstrate any significant effects the following year. Two sprays of fenitrothion applied against second generation larvae in July and August decreased numbers of over-wintering larvae, lesion development on the canes in winter, and the percentage of dead canes the following spring. In consequence, this treatment more than doubled the yield of untreated plots. A single spray of fenitrothion in June had no significant effect on subsequent lesion development but gave some control of cane death and increased yield by 55%. One spray of HCH in June had no effect. In this trial fruit yields were closely related to the extent of 'patch' lesions and less closely related to the extent of 'stripe' lesions. Both types of lesion, revealed by scraping primocanes in November, were often present on the same cane and both were decreased most by fenitrothion applied to control second generation larvae.