<p>Metabolic rate determines the amount of energy an organism needs to survive, and it is typically predicted to increase with warming up to an optimum temperature for ectothermic organisms. Once their metabolic demands have been met, organisms can use the excess energy from feeding for enhanced growth and reproduction. Experimental evidence suggests that metabolic rate may increase more with warming than energy intake, which could lead to energetic inefficiency and population decline. Downregulating metabolic rates or enhancing feeding rates after chronic exposure to warmer environments could help overcome this problem, but populations and individuals may vary in their capacity for such change. Here, we experimentally measured the temperature-dependent metabolic and feeding rates of brown trout (<i>Salmo trutta</i>) originating from one cold and two warm streams in the same geothermally heated catchment, and examined their population genetic structure. We found a consistent increase in metabolic rate with temperature for all fish, but a stronger increase in feeding rate with temperature for those originating from warm streams. This resulted in the latter exhibiting a greater energetic efficiency with increasing temperature than the fish originating from the cold stream. We detected significant genetic differentiation at neutral markers between the cold and warm streams, implying limited gene flow across the thermal or geographic gradient, and thus scope for adaptive divergence. Collectively our results point towards important variation in eco-physiology within a single catchment that has implications for population persistence in the face of warming. These results highlight the importance of considering intraspecific variation in predictive models of biological responses to climate change. They moreover emphasise how energy intake versus expenditure can be differentially thermally sensitive even at fine spatial scales.</p>