Empirical research has not supported the prediction that populations o
f terrestrial herbivorous arthropods are regulated solely by their nat
ural enemies. Instead, both natural enemies (top-down effects) and res
ources (bottom-up effects) may play important regulatory roles. This r
eview evaluates the hypothesis that higher-order predators may constra
in the top-down control of herbivore populations. Natural enemies of h
erbivorous arthropods generally are not top predators within terrestri
al food webs. Insect pathogens and entomopathogenic nematodes inhabiti
ng the soil may be attacked by diverse micro- and mesofauna. Predatory
and parasitic insects are attacked by their own suite of predators, p
arasitoids, and pathogens. The view of natural enemy ecology that has
emerged from laboratory studies, where natural enemies are often isola
ted from all elements of the biotic community except for their hosts o
r prey, may be an unreliable guide to field dynamics. Experimental wor
k suggests that interactions of biological control agents with their o
wn natural enemies can disrupt the effective control of herbivore popu
lations. Disruption has been observed experimentally in interactions o
f bacteria with bacteriophages, nematodes with nematophagous fungi, pa
rasitoids with predators, parasitoids with hyperparasitoids, and preda
tors with other predators. Higher-order predators have been little stu
died; manipulative field experiments will be especially valuable in fu
rthering our understanding of their roles in arthropod communities.