We tested the ability of consumer-resource theory to predict direct and indirect interactions among species, using an experimental system of insect herbivores and herbaceous plants. Specifically, we examined interactions among three species of grasshoppers (Melanoplus femur-rubrum, Spharagemon collare, and Phoetaliotes nebrascensis; Orthoptera, Acrididae) and herbaceous plants in experimental field cages placed over existing fertilized or unfertilized vegetation in a Minnesota old field. For the conditions inside these cages, we addressed whether (1) grasshopper diet predicted the presence of competition among grasshopper species, and (2) direct effects of grasshoppers on plants produced indirect interactions among plants, grasshoppers and soil nitrogen. Overall, M. femur-rubrum ate a greater proportion of forbs in cages, while the other two species ate primarily grasses. As expected, a pair of grasshopper species competed if they had similar diets. However, there were important exceptions that could be explained from observed indirect effects, although alternative explanations were also possible. First, all three grasshopper species significantly shifted their diets in the presence of other species, and these shifts occurred most often when competition was expected or occurred. Second, the two grassfeeding species reduced the biomass of the dominant grass (Schizachyrium scoparium) and increased available soil nitrogen and biomass of forbs. This effect may explain why the grass-feeding P. nebrascensis had a positive effect on the forb-feeding M. femur-rubrum on unfertilized plots. Overall, we show that direct effects of consumers on resources can predict competition and other important indirect interactions within a community.