The ‘fettered’ fraction of abscisic acid (ABA) that is held within the chloroplasts of unwilted bean and Commelina communis leaves is released when the leaves wilt and it is this ‘free’ ABA that is now proposed to cause the stomata to close within 2 or 3 min, well before the rise in total ABA can be detected. The large increase in ‘stress’ ABA begins 2–3 h later. The fettered ABA in a centrifuged homogenate is released by hyperosmotic solutions of mannitol (0.8 M) and NaCl (0.4 M). Dilute solution of halothane (10 mM) and colchicine (1 mM), the detergent sodium dodecyl sulphate (1 mM) the herbicide 2,4-d (0.1 mM) and dithiothreitol (0.01 mM) also caused ABA to be released. Zeatin (0.01 mM), cumene hydroperoxide (0.01 mM) and CaCl2(1 mM) had negligible effects. It was postulated that the ABA released from the chloroplasts by wilting could be the signal that initiates the synthesis of the dioxygenase and other enzymes necessary to produce the rise (up to 40-fold) in the amount of stress ABA that is seen 2–3 h later. To test this hypothesis, a solution of (+)-[2H6]ABA was vacuum infiltrated into unwilted Commelina leaves to mimic the rise in ABA caused by wilting and gas chromatography/mass spectrometry of the ABA in the extract after 3 h showed that concentrations of (+)-[2H6]ABA of up to 0.3 µM stimulated synthesis of endogenous [1H]ABA by 15-fold in the unwilted leaves. A 0.5 µM solution blocked the increase in the amount of ABA formed and also reduced the amount of ABA formed in response to a 0.8 M mannitol solution.