Biophysical and biochemical information about plant growth regulators, biomembranes and cell compartments of stressed and unstressed leaves is presented. These data are incorporated into a physiological source-sink network, which allows the calculation of phytohormone concentrations at any time in each compartment on the basis of biophysical and biochemical laws. The following results and conclusions are deduced and discussed: (i) The summarized physicochemical properties (e.g. p Ka, partition coefficient octanol: water, membrane conductance of neutral and charged phytohormone species) differ between all known phytohormones. (ii) This information is sufficient to explain experimentally observed distribution and redistribution pattern of ABA. (iii) Only cytokinins and the ethylene precursor amino-cyclopropane-carboxylic acid are distributed evenly between cell compartments, if synthesis and degradation are absent. Only under these conditions does the bulk concentration of these growth regulators in plant tissue homogenates estimate concentrations in all compartments, (iv) For other growth regulators, there are uneven compartmental concentrations depending on pH, membrane potential and anion conductance of biomembranes, even if synthesis and degradation are absent, (v) Abscisic acid is the only phytohormone which distributes almost ideally according to the anion-trap mechanism for weak acids. Calculated expected values and measurements coincide, (vi) Under diurnal illumination regimes, the same redistribution pattern of ABA for C3 and CAM plants is expected. The influence of the extreme vacuolar pH change is small because of the low ABA percentage in CAM mesophyll vacuoles (maximum 2.7 % of the total ABA mass per unit leaf area), (vii) Under drought stress, complex compartmental pH-shifts in leaves induce a complicated redistribution of ABA amongst compartments, (viii) The final accumulation of ABA in guard cell walls is up to 16.1–fold over the initial value, (ix) A 2- to 3-fold ABA accumulation in guard cell walls is sufficient to induce closure of stomata. (x) The minimum time lag until stomata start to close is 1–5 min and it depends on the stress intensity and guard cell sensitivity to ABA. (xi) The primary target membrane of ‘stress’ is the plasmalemma, not thylakoids. (xii) The effective ‘stress sensor’, which induces the proposed signal chain finally leading to stomatal closure may be located in epidermis cells. Mesophyll cells support stomatal closure only synergistically. (xiii) The direct biophysical influence of drought stress (increase of transpiration until stomata close) on the ABA concentration in guard cell walls is considerable but slow, (xiv) A stress signal from the root system in the form of an increased ABA concentration is capable of regulating the stomatal conductance, if guard cell sensitivity to ABA remains constant. The total ABA content per unit leaf area declines only within about 1 or 2 wks (aftereffect), (xv) For other phytohormones, there is no, or only a moderate, redistribution after compartmental pH changes. For biophysical reasons, only ABA is principally capable of being a ‘stress messenger’ for stomata, and evolution appears to have selected it.