Forecasting stream water quality is important for numerous aspects of resource protection and management. Fecal coliform is one of the primary indicator organisms used to assess potential pathogen contamination in drinking water supplies. Consequently, modeling the occurrence and concentration of fecal coliform is an important tool in watershed management. While many process-based, statistical, and empirical models exist for water quality prediction, artificial neural network (ANN) models are increasingly being used for forecasting of water resources variables because ANNs are often capable of modeling complex systems for which behavioral rules are either unknown or difficult to simulate. This research presents the preliminary results of an ANN model developed to predict fecal coliform concentrations in a tributary to the Wachusett Reservoir, which supplies drinking water to the metro Boston area. The model was developed using water quality parameters easily and rapidly measured in the field and standard meteorological data. In addition to assessing model performance, the effect of input data selection and preparation and ANN model architecture on model performance is assessed.