To determine whether to use single or multiple predator species for biological pest control requires manipulative field experiments. We performed such tests in Benin (West Africa) in cassava fields infested by the cassava green mite Mononychellus tanajoa, and the cotton red mite Oligonychus gossypii. These fields also harboured the cassava apex-inhabiting predator Typhlodromalus aripo and either the leaf-inhabiting predator Amblydromalus manihoti or Euseius fustis. We manipulated predator species composition on individual plants to determine their effect on prey and predator densities. In fields with T. aripo plus A. manihoti, M. tanajoa densities were reduced by T. aripo alone or together with A. manihoti, but neither of these predators, alone or together, reduced O. gossypii densities. In fields with T. aripo plus E. fustis, T. aripo alone or together with E. fustis exerted significant control over O. gossypii, but weak control over M. tanajoa. Densities of any of the predator species were not affected by co-occurring predator species, suggesting a minor role for intraguild predation in the field, contrary to earlier experiments on small plants in the laboratory. We conclude that (1) T. aripo is the most effective predator species in suppressing M. tanajoa, (2) two predator species, T. aripo and E. fustis, are needed to suppress O. gossypii, and (3) predator species together on the same plant do not negatively affect each other nor the extent to which they control their prey. We argue that intraguild predation is reduced due to partial niche separation among predator species.