For most angiosperms, flowers are critical to reproduction because they increase rates of outcrossing. Flowers are highly variable in numerous traits, including size, shape, and color. Most of this variation is thought to have arisen due to selection by pollinators. Yet, non-pollinator selection is increasingly being recognized as contributing to floral trait evolution. One such non-pollinator agent of selection that often opposes pollinator selection includes the physiological and resource costs of producing and maintaining flowers. Yet, our understanding of the physiological function of flowers remains limited. Here I argue for a comparative, physiological framework for studying flowers and highlight recent work elucidating some of the basic physiological structure-function relationships of flowers and apply an energy balance model to show how two pollination traits (flower size and flower color) can interact with physiological traits. This modeling shows that (1) physiologically meaningful estimates of flower size can be easily gleaned from available floras, (2) the range of flower size most common in the California flora is the range in which complex, non-linear dynamics in flower energy balance occur, and (3) that intraspecific variation in flower color can, by these conservative estimates, cause large (up to 8°C) changes in flower temperature. The strong interaction between pollination traits (flower size and color) and physiological traits (surface conductance to water vapor) suggests that a more unified framework for understanding the evolution of floral form and function would include physiological traits that represent the costs of flower production and maintenance.