The sensitivity of water stress indicators to changing moisture availability, and their variability, determine the number of measurements that should be taken in order to represent properly plant water status in a certain orchard. In the present study we examined the sensitivity and variability of maximum daily trunk shrinkage, midday stem water potential, and daily transpiration rate in their responses to withholding irrigation from field-grown drip-irrigated `Golden delicious' apple trees in a commercial orchard. Irrigation was withheld from the stressed trees for 17 days starting in mid-July, and the control trees were irrigated daily at 100% of the “Class A” pan evaporation rate. The courses of daily transpiration rate, maximum trunk shrinkage, and midday stem water potential before and 10 days after the drying period were similar in the control and the stressed trees. Highly significant differences between the stressed and the control trees in their midday stem water potentials were apparent from the early stages of the stress period. Daily transpiration rate and maximum daily shrinkage were more variable than midday stem water potential, and differences between treatments became significant only after measurements were expressed relative to the initial values before irrigation was witheld. Differences between treatments (as percentages of the values obtained for the control trees) increased after irrigation stopped where these differences were greatest for maximum daily shrinkage, which reached 90%; moderate for stem water potential (60%); and least for daily transpiration rate, for which the differences remained below 20%. Our data show that the choice of a certain water stress indicator should be based on both the sensitivity to changing moisture availability and the degree of variability. Possible reasons for the different sensitivity to moisture availability and the different variability between the water stress indicators under study are discussed.