1. Two field experiments were carried out to examine the role of patch size, host density, and complexity of the surrounding habitat, on the foraging behaviour of the parasitoid wasp Cotesia glomerata in the field. 2. First, released parasitoids were recaptured on patches of one or four Brassica nigra plants, each containing 10 hosts that were placed in a mown grassland area. Recaptures of females were higher than males, and males and females aggregated at patches with four plants. 3. In experiment 2, plants containing 0, 5 or 10 hosts were placed in unmown grassland plots that differed in plant species composition, on bare soil, and on mown grassland. Very low numbers of parasitoids were recaptured in the vegetated plots, while high numbers of parasitoids were recaptured on plants placed on bare soil or in mown grassland. Recaptures were higher on plants on bare soil than on mown grassland, and highest on plants containing 10 hosts. The host density effect was significantly more apparent in mown grassland than on bare soil. 4. Cotesia glomerata responds in an aggregative way to host density in the field. However, host location success is determined mostly by habitat characteristics, and stronger host or host-plant cues are required when habitat complexity increases.