Recent meta-analyses have shown that the effects of climate change are detectable and significant in their magnitude, but these studies have emphasized the utility of looking for large-scale patterns without necessarily understanding the mechanisms underlying these changes. Using a series of case studies, we explore the potential pitfalls when one fails to incorporate aspects of physiological performance when predicting the consequences of climate change on biotic communities. We argue that by considering the mechanistic details of physiological performance within the context of biophysical ecology (engineering methods of heat, mass and momentum exchange applied to biological systems), such approaches will be better poised to predict where and when the impacts of climate change will most likely occur.