The sensitivity of French prune (Prunus domestica L. syn. `Petite d'Agen') to water deprivation at various fruit growth stages was studied over 3 years in a drip-irrigated orchard. The soil was a poorly drained Rocklin fine sandy loam with a hardpan that varied from 4.75 to I m from the surface at the northern end of the orchard (shallow soil condition) to no hardpan apparent to 2 m below the surface at the southern end of the orchard (deep soil condition). Water deprivation during a) the first exponential phase of fruit growth or stage I, b) lag phase of fruit growth or stage II, c) first half of stage II, d) second half of stage II, e) second exponential fruit growth phase or stage III, and f) postharvest was compared to a fully watered control. Water deprivation caused the most severe reduction in tree water status when it was imposed over longer periods of time and during periods of high evaporative demand and also had mm-e severe effects under shallow soil conditions. Compared to the control treatment, deprivation during all of stage II (the most severe deprivation treatment) was associated with increased Ilowering, reduced fruit hydration ratio, and smaller fruit size under all soil conditions. Under deep soil conditions, deprivation during all of stage II resulted in increased return bloom, which was reflected in higher fruit loads and dry t-ha-' fruit yield. However, under shallow soil conditions, even though return bloom was increased with this treatment, fruit loads and dry t·ha-1 fruit yields were the lowest of all treatments. These differences in treatment effects in shallow vs. deep soil conditions were most likely the result of increased fruit drop, which occurred under shallow soil conditions as a result of rapid onset and increased severity ofstress. Treatments that had parallel effects in shallow and deep soil conditions resulted in statistically significant overall treatment effects, while those that had opposing effects in shallow vs. deep soil conditions did not show significant overall treatment effects. Substantial alternate hearing occurred, and, in general, dry fruit yields above ≈9 dry t·ha-1 resulted in a decrease in fruit load the following year, while loads below this value showed a subsequent increase. Based on a separate estimate of the theoretically stable value for each treatment, all deprivation treatments resulted in a higher sustainable fruit load compared to the fully irrigated control. This suggests that, for the purpose of prune fruit production, there may be an optimal level of tree water stress.