The development of practical field criteria for timing water supply is required to improve crop productivity. Among all the proposed methods for measuring plant water status, the predawn leaf water potential, PLWP, seems to be the most simple physiological indicator. For the majority of annual crops, this indicator is related to relative evapotranspiration. A similar relationship has been shown between relative transpiration and PLWP for young trees in containers or in orchard. However exceptions can occur in case of heterogeneous distribution of soil water content. To address this question, walnut trees were submitted to two experimental water content distributions (homogeneous and heterogeneous) and for two conditions (orchard and container). For heterogeneous water content distribution, we used split root system in container with 2 compartments (20% and 80% of the total available soil volume respectively) and with irrigation restricted to the small compartment only and, in orchard, we used a daily irrigation with small quantity of water (20% of water supply of the control trees). PLWP and sap flow were measured in all cases (homogeneous and heterogeneous). The experiment shows in heterogeneous conditions, that PLWP of the control and of the treatment does not differ significantly when the sap flow pattern attests water stress for the treatment. We therefore conclude that under this particular condition of heterogeneous soil water content distribution, the use of this indicator is very questionable.