Estimates of climate impacts often show that extreme temperatures have large effects on a range of outcomes. To document such non-linearities, many studies count days in different temperature ranges, or “bins,” and consider how exposure to distinct bins affects outcomes. We show that these non-linear approaches to measuring exposure to changing temperatures can generate spurious results. Specifically, global warming induces mechanical trends in extreme temperature exposure that correlate strongly with a location’s baseline temperature. Substantial bias emerges if trends in the outcome variable also correlate with baseline temperature for any reason.We demonstrate this problem, which we term binning bias, theoretically, in simulations, and with real outcome variables. We then develop solutions. In applications,some results in the literature are unaffected by these corrections, while other results change substantially.