The analysis of the electrophysiological activity of plants permits a real-time information of the plant status (e.g. light availability and water stress). However, even though it is clear that the role of the electrical signals in plant is crucial, especially in processes involving the propagation of rapid signals, a systematic approach for the interpretation of the electrical patterns is still missing. In this work a multi-electrodes approach has been applied to study the electrical signals in olive trees plants subjected to three different level of water stress. In particular, by using specific water irrigation regimes, a control group, a mid/mild stressed group and a high stressed group have been monitored and subsequently subjected to a long period of prolonged drought stress. Physiological parameters and electrical activity have been continuously monitored for the whole experiment to highlight any correlation between the electric signal and the water stress. Our results showed that it has been possible to differentiate the electric signals related to drought conditions of different intensity (i.e. control, mild and high). In particular we have found that the average daily relative electrical resistance change, the opposite of the electrical conductance, is directly related to the drought stress whilst the signal variance increases during the period of main water stress. Additionally, a proposed signal classification system has been successfully able to detect the absence/presence of stress and to effectively recognize daily class samples (93 % control, 76 % mild and 80 % high). The set-up could provide a useful tool for monitoring water conditions in plants and has several potential applications for sensor and automatic system in greenhouse or field able to monitor directly the plant water status.