Ricardian (hedonic) analyses of the impact of climate change on farmland values typically assume additively separable effects of temperature and precipitation. Model estimation is implemented on data aggregated across counties or large regions. We investigate the potential bias induced by such approaches by using a large panel of farm-level data. Consistent with the literature on plant physiology, we observe significant non-linear interaction effects, with more abundant precipitation acting as a mitigating factor for increased heat stress. This interaction disappears when the same data is aggregated in the conventional manner, leading to predictions of climate change impacts which are significantly distorted.