Mechanistic forecasts of how species will respond to climate change are highly desired but difficult to achieve. Because processes at different scales are explicit in such models, careful assessments of their predictive abilities can provide valuable insights that will be relevant to functionally similar species. However, there are surprisingly few comprehensive field tests of mechanistic niche models in the literature. We applied a general, thermodynamically grounded modeling framework to determine the fundamental niche of an extremely well‐studied herbivorous ectotherm, the sleepy lizard Tiliqua rugosa. We then compared the model predictions with detailed long‐term field observations that included sub‐hourly data on microclimate, activity levels, home ranges, and body temperatures as well as annual to decadal patterns of body condition and growth. Body temperature predictions inferred from gridded climatic data were within 10% of empirically observed values and explained >70% of observed daytime activity patterns across all lizards. However, some periods of activity restriction were explained by predicted desiccation level rather than by temperature, and metabolically driven activity requirements were much lower than potential activity time. Decadal trajectories of field growth and body condition could also be explained to within 10% of observed values, with the variance in trajectories being attributable to whether individuals had access to permanent water. Continent‐wide applications of the model partly captured the inland distribution limit, but only after accounting for water limitations. Predicted changes in habitat suitability under six climate change scenarios were generally positive within the species’ current range, but varied strongly with predicted rainfall. Temperature is regarded as the major factor that will restrict the distribution and abundance of lizards and other terrestrial ectotherms under climate change. Yet our findings show how water can be more important than temperature in constraining the activity, habitat requirements, and distribution limits of terrestrial ectotherms. Our results demonstrate the feasibility of first‐principles computation of the climatic limits on terrestrial animals from gridded environmental data, providing a coherent picture for how species will respond to climate change at different scales of space and time.