Laboratory and field‐cage tests were done to determine how undersowing brassica plants (Brassica oleraceae L. and B. rapa L.) (Cruciferae) with subterranean clover (Trifolium subterraneum L.) (Papilionaceae) affected host‐plant selection by eight pest insect species of brassica crops. The pest species tested were Pieris rapae (Lepidoptera: Pieridae) (the small white butterfly), Pieris brassicae (Lepidoptera: Pieridae) (the large white butterfly), Delia radicum (Diptera: Anthomyiidae) (the cabbage root fly), Phaedon cochleariae (Coleoptera: Chrysomelidae) (the mustard beetle), Plutella xylostella (Lepidoptera: Yponomeutidae) (the diamond‐back moth), Evergestis forficalis (Lepidoptera: Pyralidae) (the garden‐pebble moth), Mamestra brassicae (Lepidoptera: Noctuidae) (the cabbage moth) and Brevicoryne brassicae (Hemiptera: Aphididae) (the cabbage aphid). In all tests, except two in which the brassica plants were about three times as high as the clover background, 39%–100% fewer of the pest insect stage monitored were found on host plants presented in clover than on those presented in bare soil. Contrary to claims supporting the ‘enemies hypothesis’, differences in colonization alone appeared sufficient to account for the lower numbers of insects found when host plants are undersown with clover. To be effective in reducing plant colonization, the clover must cover 50%, and preferably more, of the vertical profile of the crop plants. As clover used as an undersown crop often has to be cut to make it less competitive with the main brassica crop, temporal aspects of the condition of the clover during critical periods of pest activity need to be recorded carefully before concluding that undersowing does not produce the effect desired against certain pest species under field conditions. The effective clover barrier is like any other treatment, if it is not present at the appropriate time it cannot be expected to reduce pest insect numbers.