Agricultural areas in South of France have to deal with long and severe periods of drought, which can be followed by heavy rains. This kind of events tend to be more frequent because of climate change. Agroforestry systems may promote and maintain several ecosystem services such as soil C storage, and the regulation of soil erosion, soil biodiversity etc…and are thus of growing interest in agroecology. These systems enhance plant diversity compared to conventional cropping systems and may present higher resilience to climate change. The aim of our study was thus i) to assess the impact of an agroforestry system on the resistance and resilience of soil microbial communities to water stress, ii) to analyze the stoichiometric response of the soil microbial biomass to water stress by considering a spatial gradient between the tree line and the middle of the intercrop. Our hypotheses were that the resistance and resilience of microbial communities are greater on the tree line and decreases with the distance from it, and that drought stress induces microbial biomass C:N:P imbalance. Soils were sampled at the Restinclières experimental site, (South of France, nearby Montpellier). This site gathers 20-year-old walnut trees and wheat/barley/pea crop rotation agroforestry system. The soils were sampled on the tree line and at several distances perpendicular to the tree line in the pea intercrop. To test our hypotheses, we simulated two cycles of drying-rewetting disturbance on these agroforestry soils in lab conditions. We followed microbial respiration during 3 months and microbial biomass C, N and P were assessed immediately at the end of the first drying stress period (resistance), and 3, 7 and 30 days after rewetting (resilience). Based on cumulative respiration, our results indicated that the first water stress impacted faster and stronger microbial communities than the second stress. Soil samples from the tree line were impacted faster by water stress, thus leading to resistance index of microbial biomass C lower on the tree line (0.57) than in the middle of the intercrop (0.68). For resistance index related to microbial biomass N the trend was opposite and C:N ratio were significantly higher in stressed samples. After 3 months of incubation microbial biomass C,N and P values of stressed soil samples were equivalent to the control values. Resilience index values for microbial biomass C were higher on the tree line (0.76) than in the middle of the intercrop (0.34). Microbial activity was higher close to the tree line and decreased with distance from the tree. Close to the tree line, microbial communities seemed to be more sensitive to water stress, which refutes our initial hypothesis on microbial resistance. Microbial biomass C:N:P stoichiometry seemed to be affected by water stress, according to our hypothesis.