Fitness of herbivorous ectotherms is strongly influenced by temperature and diet; an interaction generally thought to be mediated via effects of temperature on metabolic rate. Using locusts as a model, we found that whereas metabolic rate increased with temperature, the rates at which protein and carbohydrate were absorbed from particular plants did not scale similarly with temperature: an interaction further complicated by body size. Consequently, the effect of temperature on life history outcomes is host plant specific; i.e. host plant ‘quality’ is temperature dependent. Locusts were able to exploit this temperature-host plant interaction to choose body temperatures so that protein and carbohydrate were differentially absorbed to redress experimentally imposed nutrient imbalances. Linking laboratory findings with host-plant and temperature choice in the field, we demonstrate that for locusts the effect of temperature on life history outcomes are complex, and host-plant choice is dependent on both the range of temperatures and host-plants available.