We studied the relief of water stress associated with fruit thinning in pear (Pyrus communis L.) trees during drought to determine what mechanisms, other than stomatal adjustment, were involved. Combinations of control irrigation (equal to crop water use less effective rainfall) and deficit irrigation (equal to 20% of control irrigation), fruit load (unthinned and thinned to 40 fruits per tree) and root pruning (pruned and unpruned) treatments were applied to pear (cv. 'Conference') trees during Stage II of fruit development. Daily patterns of midday stem water potential (Psi(stem)) and leaf conductance to water vapor (g(l)) of deficit-irrigated trees differed after fruit thinning. In response to fruit thinning, gl progressively declined with water stress until 30 days after fruit thinning and then leveled off, whereas the effects of decreased fruit load on Psi(stem) peaked 30-40 days after fruit thinning and then tended to decline. Soil water depletion was significantly correlated with fruit load during drought. Our results indicate that stomatal adjustment and the resulting soil water conservation were the factors determining the Psi(stem) response to fruit thinning. However, these factors could not explain differences in daily patterns between g(l) and Psi(stem) after fruit thinning. In all cases, effects of root pruning treatments on Psi(stem) in deficit-irrigated trees were transitory (Psi(stem) recovered from root pruning in less than 30 days), but the recovery of Psi(stem) after root pruning was faster in trees with low fruit loads. This behavior is compatible with the concept that the water balance (reflected by Psi(stem) values) was better in trees with low fruit loads compared with unthinned trees, perhaps because more carbon was available for root growth. Thus, a root growth component is hypothesized as a mechanism to explain the bimodal Psi(stem) response to fruit thinning during drought.