Network hydraulic models are widely used, but their overall accuracy is often unknown. Models are developed to give utilities better insight into system hydraulic behavior, and increasingly the ability to predict the fate and transport of chemicals. Without an accessible and consistent means of validating a given model against the system it is meant to represent, the value of those supposed benefits should be questioned. SCADA databases, though ubiquitous, are underused data sources for this type of task. Integrating a network model with a measurement database would offer professionals the ability to assess the model's assumptions in an automated fashion by leveraging enormous amounts of data. EPANET-RTX, the real-time extension to the EPANET toolkit, allows this integration by communicating between SCADA, a hydraulic model (EPANET), a demand estimator, and a results database. The extension is described here and verified through application to a synthetic dataset. The extension is then used to link a live SCADA database to a utility's network model, and assess that model by comparing the simulated tank levels to the utility's database of historic tank levels. EPANET-RTX updates the model automatically to reflect historical pump status and boundary conditions, and a few rudimentary demand estimation algorithms are applied. The model's performance is evaluated by computing the average difference in modeled and measured values across all of the system's tanks. Although the network model investigated here was recently developed by consultants to the water utility, a preliminary analysis based on one week of historical data indicates that it may perform poorly in terms of predicting tank levels when connected to real pumping and control data. In the authors view this has less to do with the skill of the consultants, than the need to integrate hydraulic models with richer data sources that can allow adaptation of key model parameters to the continuously changing operation environment. Indeed, implementing very simple demand estimation and scaling algorithms improves model performance, which makes a strong case for the need to develop tools that leverage SCADA data for modeling purposes.