To manage drinking water more efficiently, non‐revenue water should be reduced, which can only been done after analyzing non‐revenue water. The current process of analyzing the performance of non‐revenue water or its components should be more efficient in time. To take a step forward to a more time sufficient approach it is proposed, in cooperation with Royal HaskoningDHV, to examine the relationships between the performance of non‐revenue water or its components and the variables that influence or are influenced by the performance. The performance of non‐revenue water is expressed in liters/connection/day. The variables included in the model are the connection density, the pressure, the night‐peak factor, the number of connections and the sector type. The relationships are shown in structural equation models created using the path analysis; the equations evolving from the models are analyzed using the multiple backward elimination analysis. An important relationship which is proven by the analysis is the influence of the performance of the real losses on the night peak factor. The model also shows that the pressure does not influence the performance of the real losses. The results from the 95% confident intervals, the variance and shrinkage in the model show that the structural equations are too inaccurate to be used for predictive purposes.