The stomatal conductance of several anisohydric plant species, including field-grown sunflower, frequently correlates with leaf water potential ([psi]1), suggesting that chemical messages travelling from roots to shoots may not play an important role in stomatal control. We have performed a series of experiments in which evaporative demand, soil water status and ABA origin (endogenous or artificial) were varied in order to analyse stomatal control. Sunflower plants were subjected to a range of soil water potentials under contrasting air vapour pressure deficits (VPD, from 0.5 to 2.5 kPa) in the field, in the glasshouse or in a humid chamber. Sunflower plants were also fed through the xylem with varying concentrations of artificial ABA, in the glasshouse and in the field. Finally, detached leaves were fed directly with varying concentrations of ABA under three contrasting VPDs. A unique relationship between stomatal conductance (gs) and the concentration of ABA in the xylem sap (xylem [ABA]) was observed in all cases. In contrast, the relationship between [psi]1 and gs varied substantially among experiments. Its slope was positive for droughted plants and negative for ABA-fed whole plants or detached leaves, and also varied appreciably with air VPD. All observed relationships could be modelled on the basis of the assumption that had no controlling effect on gs. We conclude that stomatal control depended only on the concentration of ABA in the xylem sap, and that [psi]1 was controlled by water flux through the plant (itself con-trolled by stomatal conductance). The possibility is also raised that differences in stomatal 'strategy' between isohydric plants (such as maize, where daytime [psi]1 does not vary appreciably with soil water status) and anisohydric plants (such as sunflower) may be accounted for by the degree of influence of [psi]1 on stomatal control, for a given level of xylem [ABA]. We propose that statistical relationships between [psi]1 and gs are only ob