The increased utilization of non-native tree species in monocultures or tree mixtures has repeatedly been shown to have negative consequences for the abundance and biomass of natural enemies. However, it is often unknown whether these effects translate into lower top-down control, and if so, whether these effects are driven by tree diversity and identity or by factors not strictly attributed to tree identity (e.g. light conditions or vegetation structure). Here, we used model caterpillars to study the effect of non-native Douglas fir (Pseudotsuga menziesii) on pest control in Central European forests. Model caterpillars were set at two strata (ground and herb layer) in 40 plots of five forest stand types. Monocultures of European beech (Fagus sylvatica), Douglas fir and Norway spruce (Picea abies) and two-species mixtures of European beech and each of the conifers were studied. We analyzed plot-level attack rates (total, arthropod, bird, and mammal attack rates) in relation to stand type, stand composition and microhabitat variability. Against expectations, stands with a higher proportion of Douglas fir had higher total and arthropod attack rates. However, this effect was lessened in stands with higher tree species diversity. Indirect effects of Douglas fir were stronger than direct tree identity and proportion effects, with both increased light availability and understory vegetation complexity promoting attack rates. Therefore, promoting vegetation complexity and light availability by thinning, or planting more open Douglas fir stands with a lower stem density could help to promote pest control in forest stands of non-native tree species.