Abscisic acid (ABA) is a stress signal that is transported from the root system to leaves, and induces stomatal closure before water relations of the leaves are affected by soil drying. Xylem vessels are in direct contact with the leaf apoplasm, the only leaf compartment that is directly connected with the primary site of ABA action, the outer surface of the guard cell plasma membrane (Hartung 1983). ABA distributes among the leaf compartments according to the anion trap concept and the Henderson-Hasselbalch equation, with the free acid as the permeating and the anion as the nearly non-permeating molecular species. Applying this concept, a flattening of the intracellular pH gradients increases the apoplastic ABA concentration. Indeed, stress increases the apoplastic pH (Hartung et al. 1988) and decreases slightly the cytosolic pH. The validity of this concept has been shown repeatedly and was confirmed by a mathematical leaf model (Slovik et al. 1992). It is appropriate to ask whether these mechanisms also contribute to ABA compartmentation and redistribution in the root system. Therefore, we have incorporated compartmental pH values of unstressed and stressed root cells, the permeability coefficients of root membranes for ABA and anatomical data into a mathematical model, similar to that of Slovik et al. (1992). The simulation shows that ABA redistribution in roots caused by changing pH gradients can account for up to a 2 to 3-fold accumulation of ABA in the xylem sap of stressed plants. The model also predicts that the pH gradient across the cortical plasma membrane has the most distinct effects on redistribution of ABA into the xylem sap of stressed plants and, additionally, that the ABA concentration in the rhizospheric aqueous solution can play an important role in root-to-shoot signalling.