Previous studies of insect-plant interactions have produced the contradictory Plant Stress Hypothesis (that stressed plants are more suitable hosts for phytophages) and Plant Vigor Hypothesis (that vigorous plants are more suitable hosts for phytophages). However, experimental studies of phytophage responses to host stress have involved only one, or a few, related phytophagous species, not whole communities of organisms associated with a particular plant species. We evaluated responses of various arthropods associated with creosotebush Larrea tridentata to manipulated water availability and plant stress in southern New Mexico during 1990-1991. Of 44 arthropod groups (taxa or functional groups) evaluated in our study only two taxa (including a lepidopteran folivore) showed significant negative response to water availability, thereby supporting the Plant Stress Hypothesis. Ten taxa (including eight phytophages) responded positively to water availability, supporting the Plant Vigor Hypothesis. One phytophage showed a nonlinear response, supporting neither hypothesis. Detrended Correspondence Analysis significantly distinguished the arthropod community on water-deprived shrubs from the communities on watered shrubs. The variation in responses among phytophagous insects on creosotebush indicated that the effect of plant water stress likely reflects the choice of phytophage, and perhaps the host plant, being studied. Therefore, neither the Plant Stress Hypothesis nor the Plant Vigor Hypothesis can explain responses of all phytophages on a particular plant species.