Security is an important concern for the water industry nowadays. It is particularly important to have tools that are able to determine the source of contamination once an alarm is raised by the sensor detection system. The objective of this paper is the development of such a model to locate potential sources of contamination. The model solution uses a two-step method. First, a linear source-response transport equation is built based on a backtracking algorithm applied to sensor binary answers. It is followed by a minimum relative entropy method or a least squares method coupled with a Tikhonov regularization that yields the probability density function of the different potential sources found at the first step. The model has been validated on small networks and applied to a large network under single and multiple contamination scenarios. This research is part of the European project on water network security named SecurEau.