Predicting food web responses to climate change can be difficult because of the potentially complex interplay between co-occurring climate variables and multiple interacting species across trophic levels. The large majority of research in this field has focused on understanding the effects of single climate variables on species at one or two trophic levels, implicitly assuming that simultaneous shifts across multiple climate variables will have additive effects on food web dynamics. We constructed a tri-trophic food web model and varied temperature, CO2, and water availability both alone and in concert to test this assumption. We found that population biomass does indeed respond additively across trophic levels when temperature, CO2, and water availability all increase simultaneously to moderate levels; however, if water availability decreases, like in a drought scenario, all three trophic levels respond antagonistically. We also found that interaction effect magnitude is highly dependent on temperature and water availability. Decreases in water availability led to 54–74% declines in population biomass across trophic levels when temperatures were within normal organismal operating ranges, but dry conditions coupled with high temperatures led to the extinction of the highest trophic level. Our results suggest that studying simplified versions of climate change and food webs will not be sufficient to predict the responses of real ecological systems. Therefore climate change ecology experiments and models must incorporate more complexity into their structure.