Mixed forests are thought to be less prone to pest insect damage than monocultures. This may result from reduced host availability (i.e., density effect) or from non-host trees reducing the physical or chemical apparency of host trees (i.e., associational resistance, AR). However, associational and density effects are often confounded in mixed forests. We aimed to disentangle their relative contribution to attacks of pine trees by a specialist pest, the pine processionary moth (PPM, Thaumetopoea pityocampa). We assessed pine infestation by PPM by counting the number of winter nests during three consecutive years along an experimental gradient of pine density in presence or absence of a fast growing species, namely birch. The total number of PPM nests per plot increased with pine density (maximum in high density monocultures), while the proportion of attacked pine trees decreased along the same gradient. Birch provided associational resistance via reduced pine apparency due to their greatest higher. This mechanism occurred at two spatial scales, whenever birch was planted within pine plots or in adjacent plots. Associational resistance was stronger in dense stands, probably due to reduced distance between pines and neighboring birches. But AR faded with time, pines becoming taller than birches, making density effects preeminent over apparency effects. Our findings suggest that mixing tree species to trigger resistance to pest insects requires taking into account the relative growth rate of associated species together with the relative proportion of associated species, both within and between stands. © 2016 Elsevier B.V.< Réduire