Experiments were done in fields of winter wheat at North Farm, West Sussex and at Rothamsted [England, UK] in 1978 to investigate the effect of predation by carabid beetles and other polyphagous predators on populations of cereal aphids. Predator populations were manipulated by pitfall trapping within plots surrounded by polythene barriers and by applying an insecticide (fonofos) within such plots. Plots were set up in March, April, May and June, and the development of aphid populations within them compared with aphid populations in unenclosed control areas. The numbers of carabid beetles caught in the enclosed and fonofos-treated plots were much less than in control plots. Aphid populations were all much smaller in control plots than in those with depleted predator populations both at North Farm where beetle populations were large and at Rothamsted where populations were small. Aphid populations were much larger in plots where numbers of polyphagous predators were decreased early in the season than where this was done later. There was an inverse relationship between numbers of polyphagous predators and aphids for the North Farm experiment. There were positive relationships between aphids and aphid-specific predators and between aphids and aphid parasites. This, together with the virtual absence of the pathogenic fungus, Entomophthora spp. indicated that differences in aphid density between plots were caused by polyphagous predators rather than by other natural enemies. Comparison of results from the 2 sites suggested that among the carabid beetles Agonum dorsale (Pont.) was the most important aphid predator, but there was also evidence that polyphagous predators other than carabids decreased the cereal aphid populations.