Water distribution systems can experience high levels of leakage resulting in major financial, supply and pressure losses. Locating and repairing system leaks can drastically reduce the amount of water that is lost, as well as reduce the costs for obtaining, treating and pressurizing water distribution systems to meet current and future demands. This paper describes the step-testing network modeling approach that solves the leakage detection problem using a direct application of network modeling enhanced with graph-theoretical search algorithms and step-testing. The technique involves bracketing the test area with excessive leakage into a tight branched network with a flow meter installed on its input main. Working from the valve furthest away from the flow meter, the size of the area is systematically reduced by closing valves to cut off different pipe sections in succession (so that less and less of the test area is supplied through the meter), at the same time recording changes in flow rate at the meter and comparing with model results. The sequence of closing valves is followed working backward towards the flow meter until the meter is reached (when the flow becomes zero). A disproportionate change in flow rate between two successive steps indicates a leak in the section of pipe that was last shut off. The sequence is repeated by opening valves in reverse order. The method can effectively narrow down leaks to specific pipe segments of the distribution system. It is normally carried out at night before the morning high demand to minimize supply interruption and inconvenience to customers. An example application is used to illustrate the proposed approach. The method should prove useful to any water utility attempting to locate excessive pipe leaks in distribution systems and conserve such a precious natural resource as water. It is modeling complexity through simplicity.