U.S. agriculture is likely to be affected by climate change due to its inherent reliance on climatic inputs. An important difference among methods of climate change impact assessment is the treatment of farmer adaptation. While the cross-sectional Ricardian approach assumes that farmers have fully adapted to their current climate, panel methods assessing weather effects on profits assume farmers cannot fully adapt to idiosyncratic weather changes. Less is known, however, about the process of climate change adaptation and how farmers transition from practices adapted to a given climate to the next. This thesis posits that farmers must first perceive that climate is changing as a pre- requisite of engaging into adaptive responses. I test whether this first step in the adaptation process is occurring by exploiting the effect of random weather fluctuations on farm real estate, which reflects farmer perceptions about farm profitability. I develop a theoretical model to clarify the channels through which random weather shocks could affect farmland values, in which I consider farmers as Bayesian learners who update their priors about their mean climate based on experienced weather. I then rely on a distributed lag model to test the hypothesis. I find no evidence that weather shocks have affected the farmland market. These findings are robust to geographic and temporal subdivisions. The results suggest that farmers do not perceive recent extreme weather as indications of sizable upcoming changes in farm profitability. This may reflect the countervailing effect of agricultural prices and of government policies such as disaster payments.