This paper presents a Kriging data-based method for leak localization in Water Distribution Networks (WDNs). Given a set of sensors that provide pressure measurements in some inner nodes of the WDN, a WDN data-driven model of the pressure at each sensed node is created. Then, a pressure estimation of the remaining WDN nodes which are not sensed is provided by the Kriging interpolation method. The proposed leak localization strategy is based on the comparison of the pressure estimation in a leak scenario to the approximated pressure map in the same working conditions but in a no-leak scenario. The approximated pressure map is obtained by applying the proposed WDN data-driven model and Kriging to the actual measurements. As result of the comparison, the node that presents the maximum residual value, i.e. the maximum difference between its current and its reference pressure value, is initially pointed as a leaky node candidate. The proposed methodology is successfully applied to a real-data case study.