The natural enemies hypothesis predicts that the abundance and diversity of antagonists such as predators and parasitoids of herbivores increases with the diversity of plants, which can lead to more effective top-down control of insect herbivores. However, although the hypothesis has received large support in agricultural systems, fewer studies have been conducted in forest ecosystems and a comprehensive synthesis of previous research is still lacking. We conducted a meta-analysis of 65 publications comparing the diversity, abundance or activity of various groups of natural enemies (including birds, bats, spiders and insect parasitoids) in pure vs. mixed forest stands. We tested the effects of forest biome, natural enemy taxon and type of study (managed vs experimental forest). We found a significant positive effect of forest tree diversity on natural enemy abundance and diversity but not on their activity. The effect of tree diversity on natural enemies was stronger towards lower latitudes but was not contingent on the natural enemy taxon level. Overall, our study contributes toward a better understanding of the “natural enemies hypothesis” in forest systems and provides new insights about the mechanisms involved. Furthermore, we outline potential avenues for strengthening forest resistance to the growing threat of herbivorous insects.