Functional trait variation measured on continuous scales has helped ecologists to unravel important ecological processes. However, forest ecologists have recently moved back toward using functional groups. There are pragmatic and biological rationales for focusing on functional groups. Both of these approaches have inherent limitations including binning clearly continuous distributions, poor trait-group matching, and narrow conceptual frameworks for why groups exist and how they evolved. We believe the pragmatic use of functional groups due to data deficiencies will eventually erode. Conversely, we argue that existing conceptual frameworks for why a limited number of tree functional groups may exist is a useful, but flawed, starting point for modeling forests that can be improved through the consideration of unmeasured axes of functional variation.