Documenting trophic links in a food web has traditionally required complex exclusion experiments coupled with extraordinarily labor-intensive direct observations of predator foraging. Newer techniques such as stable isotope analysis (SIA) may facilitate relatively quick and accurate assessments of consumer feeding behavior. Ratios of N and C isotopes are thought to be useful for determining species' trophic position (e.g., 1 degrees consumer, 2 degrees consumer, or omnivore) and their original carbon source (e.g., C3 or C4 plants; terrestrial or marine nutrients). Thus far, however, applications of stable isotopes to terrestrial arthropod food webs have suggested that high taxon-specific variation may undermine the effectiveness of this method. We applied stable isotope analysis to a pear orchard food web, in which biological control of a dominant pest, pear psylla (Cacopsylla pyricola), involves primarily generalist arthropod predators with a high frequency of omnivory. We found multiple sources of isotopic variation in this food web, including differences among plant tissues; time, stage, and taxon-specific differences among herbivores (despite similar feeding modes); and high taxon-specific variation among predators (with no clear evidence of omnivory). Collectively, these multiple sources of isotopic variation blur our view of the structure of this food web. Idiosyncrasies in consumer trophic shifts make ad hoc application of SIA to even moderately complex food webs intractable. SIA may not be a generally applicable "quick and dirty" method for delineating terrestrial food web structure-not without calibration of specific consumer food trophic shifts.