We investigated the effects of plant architecture on predator–prey interactions by quantifying the behavior of green lacewing larvae on perennial grasses with divergent leaf architectures. Crested wheatgrass produces flat, broad leaves similar to those of wheat, whereas Indian ricegrass bears linear leaves that are tightly rolled inward. In the absence of prey, lacewing time budgets and residence times were similar on the two grasses, although predators tended to search longer on crested wheatgrass. On plants infested with the Russian wheat aphid, lacewing larvae dislodged, contacted, and captured significantly more aphids on Indian ricegrass than on crested wheatgrass. Comparisons between aphid-free and aphid-infested plants suggest that differences in plant architecture modified prey accessibility rather than predator movement. Aphids on seedlings and mature plants of crested wheatgrass frequently occurred in concealed locations, such as in the rolls of immature leaves or in the blade–sheath junctions of mature leaves; aphids on Indian ricegrass were more likely to feed in exposed locations. Our focal-animal observations were consistent with results from population-level experiments and suggest that short-term, behavioral studies may help predict the effectiveness of predators at larger spatial and temporal scales.