Sanitary sewer overflows (SSO) are becoming of increasing concern to utilities and regulators as they pose a health risk. However, many overflows are not easily identifiable. A methodology is presented for detecting if a disruption in the system is occurring. The approach links neural networks as a prediction tool for expected flows and control limit theory for identifying significant deviations from the expected values that suggest an SSO occurrence. Detection depends upon the magnitude of the disruption, the relative distance between the disruption and the gage, the ability to analyze large quantities of monitoring data and the capacity of the developed methodology to compare real-time data with expected conditions. Data from Pima County Wastewater Management's monitoring system is used as a case study.