A method, the comparison of flow pattern distributions (CFPD), is described in which the specific representation of flow measurements for two different time periods allows a direct, quantitative interpretation of changes in the pattern. Two types of changes can be distinguished. The first is changes from one period to the next in demand consistent with the existing pattern, e.g. due to changing weather or changes in the population size. The second type is inconsistent changes which may be due to increased leakage. The method is successfully applied to drinking water distribution systems of different sizes and characteristics. Being data driven, it is independent of model assumptions and therefore insensitive to uncertainties therein which may hinder some other leakage determination methods. Because it is simple to implement and apply but nevertheless powerful in distinguishing between consistent and inconsistent changes in water demand, the method provides water companies with a way to constantly monitor their networks for possible changes in customer demand and the possible occurrence of new leakages and also check archived data for similar changes. This could render additional information about customer behavior and the evolving condition of the network from data which is usually readily available at water companies.