This paper presents a report on the implementation of an artificial neural network (ANN) that is to be used to forecast the daily peak load at the Pennsylvania -New Jersey -Maryland Interconnect Association (PJM). This paper examines the process of determining the appropriate ANN inputs and architectures and compares the results to the current mainframe-based pattern matching program. It is unique in that it uses real rather than simulated data to compare the ANN performance and also because it is looking at peak load forecasting for a large network of generating plan