Tomato shoots that had been (a) fed (�)-[2H9]abscisic aldehyde via the xylem or (b) fed H218O together with (�)-[2H9]abscisic aldehyde via the xylem or (c) exposed to 18O2 and fed (�)-[2H9]abscisic aldehyde, were then wilted. The abscisic acid present was isolated, methylated and resolved into (+)- and (-)- methyl abscisate. These methyl abscisate samples were then examined by negative ion chemical ionisation (methane) gas chromatography/mass spectrometry. The undeuteriated (+)-abscisic acid contained no 180 from H218O but did contain one 18O from 18O2. No 18O from either of these sources was present in the undeuteriated (-)-abscisic acid. It was not possible to discount the xanthophyll hypothesis for the origin of stress-induced abscisic acid on the basis of these experiments. Both (+)- and (-)- multiply deuteriated abscisic acid contained one and two 18O atoms from H218O but none from 18O2. It is postulated that this multiply deuteriated (�)-abscisic acid is formed by a separate enzyme system from that which forms endogenous stress-induced (+)-abscisic acid. On the basis of the low incor- poration of abscisic aldehyde into abscisic acid, it is suggested that the endogenous precursor of stress- induced abscisic acid is an as yet unidentified structure and that abscisic aldehyde competes with it.