It was shown over 40 years ago that plants maximize carbon gain for a given rate of water loss if stomatal conductance, gs , varies in response to external and internal conditions such that the marginal carbon revenue of water, ∂A/∂E, remains constant over time. This theory has long held promise for understanding the physiological ecology of water use and for informing models of plant-atmosphere interactions. Full realization of this potential hinges on three questions: (i) Are analytical approximations adequate for applying the theory at diurnal time scales? (ii) At what time scale is it realistic and appropriate to apply the theory? (iii) How should gs vary to maximize growth over long time scales? We review the current state of understanding for each of these questions and describe future research frontiers. In particular, we show that analytical solutions represent the theory quite poorly, especially when boundary layer or mesophyll resistances are significant; that diurnal variations in hydraulic conductance may help or hinder maintenance of ∂A/∂E, and the matter requires further study; and that optimal diurnal responses are distinct from optimal long-term variations in gs , which emerge from optimal shifts in carbon partitioning at the whole-plant scale.