Numerous studies conducted in agro-ecosystems support the enemies hypothesis, which states that predators and parasites are more efficient in controlling pest densities in polycultures than in monocultures. Few similar studies, however, have been conducted in forest ecosystems, and we do not yet have evidence as to whether the enemies hypothesis holds true in forests. In a 2-year study, we investigated whether the survival of autumnal moth ( Epirrita autumnata) larvae and pupae differs between silver birch monocultures and two-species mixtures of birch with black alder, Norway spruce and Scots pine. We placed young larvae on birch saplings and monitored their survival until the end of the larval period, when we checked whether they had been parasitized. After the larvae had pupated, pupal survival was tested in a field trial. In 2002, the larvae disappeared earlier and their overall survival was lower in birch-pine mixtures than in other stand types. In 2003, survival probability was lowest in birch-pine stands only during the first week and there were no differences between stands in overall survival. Larval parasitism was not affected by tree species composition. Pupal weight and pupal survival were likewise not affected by stand type. Among the predators, wood ants were more abundant on birches growing in birch-pine mixtures than in other stand types probably because colonies of myrmecophilic aphids were common on pines. In contrast, spider numbers did not differ between stand types. Ant exclusion by means of a glue ring around the birch trunk increased larval survival, indicating that ants are important predators of the autumnal moth larvae; differences in larval survival between stands are probably due to differential ant predation. Our results provide only partial support for the enemies hypothesis, and suggest that it is both tree species composition and species diversity which affect herbivore survival and predation.