Because of regulations at the whole-plant level, cells of droughted pl
ants do not necessarily experience dehydration. In fact, recent data s
uggest that they usually do not in the range of water deficits compati
ble with agriculture. In this range, leaf water potential and cell tur
gor are frequently maintained at high values in spite of decreasing so
il water status. As a consequence, 'water stress' cannot be defined by
plant water status, except when very rapid and severe water deficits
cause catastrophic events such as xylem embolism or severe cell dehydr
ation. In water deficits compatible with agricultural situations, plan
t water status is in many cases tightly controlled by the plant. 'Wate
r stress' should then be defined by water statuses at plant boundaries
, i.e. soil and air. Partial maintenance of plant water status under w
ater deficit is allowed by controls of stomatal conductance, root and
leaf expansions and leaf senescence. These processes involve both chem
ical and hydraulic signallings from roots. All these controls tend to
reduce transpiration (stomatal closure, reduced leaf growth or leaf se
nescence) or to increase water uptake (maintenance of root growth or i
ncrease in root/shoot ratio). During relatively mild stresses, the rol
e of abscisic acid, frequently considered as a 'stress hormone', in fa
ct appears to be to avoid dehydration at the cellular level. In the ca
ses described here, selecting plants for better resistance to cell deh
ydration may not be the best strategy for improving 'drought resistanc
e' of crops.