Summary1. Functional and numerical responses are key components in the selection of predators for biological control. We examined the relevance of these responses for predicting the effectiveness of generalist predators, taking into account effects of alternative prey and multiple predators. Our system involved two acarine predators of two prey species on apple. Responses were measured on leaf discs in the laboratory, and predictions assessed on small potted apple trees. In particular, we tested three hypotheses. 2. Hypothesis I: the species with the higher predatory responses will be more effective in limiting prey populations. Neither predator had a consistently higher functional response, which depended on prey stage and species. Amblyseius fallacis had a (approximately two times) higher ovipositional response than Typhlodromus pyri. We therefore hypothesized that A. fallacis would be more effective in controlling prey. No evidence was found to support this hypothesis. 3. Hypothesis II: alternative prey reduce the functional response of predators to target prey. Alternative prey did or did not reduce the functional response to target prey, depending on the predator, stage and species of alternative prey. The only consistent trend for both predators was that the predation of Panonychus ulmi deutonymphs was reduced when Tetranychus urticae was present. 4. Hypothesis III: the predator species with the highest mean ovipositional response will out-compete the other predator species. The number of predators observed in the mixed predator treatments depended on the prey composition. Although A. fallacis had a higher ovipositional response, it was never more abundant. Intraguild predation probably played a role in determining predator abundance, although prey composition altered intraguild effects. 5. These results highlight the complex nature of predator and prey interactions. Because of these interactions, functional and numerical assays on single predatorsingle prey systems in simplified laboratory environments do not allow predictions of the growth of mixed populations in realistic habitats, or of the effectiveness of predators as biological control agents in the field.