1. Homeostatic responses of animals to environmentally-induced changes in nutrient requirements provide a powerful basis for predictive ecological models, and yet such responses are virtually unstudied in the wild. 2. We tested for macronutrient-specific compensatory feeding responses by free-ranging golden snub-nosed monkeys (Rhinopithecus roxellana) inhabiting high altitude temperate forests where they experience a substantial difference in ambient temperature in cold winters vs. warmer springs. The monkeys had free access to natural foods throughout the year, and to ensure that any seasonal differences in nutrient intake were due to homeostatic compensation and not constraints on food availability, we studied the monkeys during periods in which they were provisioned with the same amount of supplementary foods in winter and spring. 3. Thermoregulatory energy costs in winter and spring were calculated using partitional calorimetric estimations of convective and radiative heat loss obtained from thermal imaging of free-ranging monkeys in situ. Daily nutrient intakes were measured using continuous focal follows (average 6.9 h/day) of free-ranging individuals (27 in spring and 28 in winter). 4. We used a nutritional geometry framework to integrate these data and test three predictions: i. In order to remain thermoneutral (balance heat loss with heat expenditure), golden snub-nosed monkeys decrease daily energy consumption during the spring compared to winter, ii. Decreased energy intake is accomplished specifically by reducing intake of the primary energetic nutrients, carbohydrate and lipid, relative to protein, and iii. The seasonal reduction in ingested fat and carbohydrate calories will quantitatively match the reduction in thermoregulatory costs in spring compared with winter. 5. Our results showed that energy intake in spring was reduced to almost half (55%) of that in winter. As predicted, this was achieved by specifically reducing fat and carbohydrate consumption with protein intake unchanged, by a quantity (326 kJ/mbm) that almost exactly matched the seasonal difference in the daily energetic costs of thermoregulation (329 kJ/mbm). 6. This is the first study to test for a match between nutrient-specific homeostatic compensation and environmentally-induced perturbations in nutrient requirements in free-ranging animals, and underpins the potential for the homeostasis framework to provide predictive power to ecological models.