Current ecosystem models do a poor job of predicting plant drought responses. In large part this is because they incorporate poor hypotheses for how stomatal regulation of water loss and carbon uptake responds to water deficits in soil and atmosphere. A pragmatic solution is to assume that canopy water loss is regulated in coordination with canopy water supply. Water supply is readily quantified (and generalized) by the interesting physics of liquid flow through the porous media of soil and plant xylem. Cavitation of the metastable liquid phase places defined limits on water supply. By logically assuming that stomatal regulation operates in proportion to the need to avoid cavitation, the full set of stomatal responses to soil moisture and atmospheric vapor pressure deficit is efficiently predicted. Initial tests of the theory are promising, suggesting that in simplified form it can find useful application in improving ecosystem models.