Increased regulation of chemical pesticides and rapid evolution of pesticide resistance have increased calls for sustainable pest management. Biological control offers sustainable pest suppression, partly because evolution of resistance to predators and parasitoids is prevented by several factors (e.g., spatial or temporal refuges from attacks, reciprocal evolution by control agents, and contrasting selection pressures from other enemy species). However, evolution of resistance may become more probable as agricultural intensification reduces the availability of refuges and diversity of enemy species, or if control agents have genetic barriers to evolution. Here we use 21 y of field data from 196 sites across New Zealand to show that parasitism of a key pasture pest (Listronotus bonariensis; Argentine stem weevil) by an introduced parasitoid (Microctonus hyperodae) was initially nationally successful but then declined by 44% (leading to pasture damage of c. 160 million New Zealand dollars per annum). This decline was not attributable to parasitoid numbers released, elevation, or local climatic variables at sample locations. Rather, in all locations the decline began 7 y (14 host generations) following parasitoid introduction, despite releases being staggered across locations in different years. Finally, we demonstrate experimentally that declining parasitism rates occurred in ryegrass Lolium perenne, which is grown nationwide in high-intensity was significantly less than in adjacent plots of a less-common pasture grass (Lolium multiflorum), indicating that resistance to parasitism is host plant-dependent. We conclude that low plant and enemy biodiversity in intensive large-scale agriculture may facilitate the evolution of host resistance by pests and threaten the long-term viability of biological control.attack rates | GAMM | invasive species | meta-analysis | natural enemy