The goal of this study was to investigate the impacts of skeletonization on the accuracy of modeling of a water distribution system. The four levels of skeletonization investigated in the study included: 1) a gross level representation based primarily on a set of rules specifying pipe sizes to include; 2) a slightly more detailed level of skeletonization using engineering judgment to expand the first level by inclusion of a few more loops; 3) a significantly more detailed level of skeletonization; and 4) inclusion of essentially all pipes. Throughout the analyses performed, the accuracy of the models, as measured by the difference in results as compared to the full system representation, increased with inclusion of more pipes. There was, however, a fairly substantial increase in accuracy between the first and second levels of skeletonization. A second important conclusion is that the impacts of skeletonization depend upon the type of analysis being performed. In general, water quality analyses dependent upon accurate representations of flow paths and travel times are more sensitive to level of skeletonization than are more standard hydraulic analyses. Additionally, the errors introduced by skeletonization are more acute under more stressful hydraulic conditions.