An attempt is made to forecast electric load using neural networks. Neural networks represent a pattern or load shape and, in reality perform a pattern recognition function. This pattern is based on training cases provided to the network. A strategy that uses a minimum distance measurement to identify the appropriate historical patterns of load and temperature readings is used to estimate the network weights. Most of the electrical power systems in India exhibit energy shortages. An optimization algorithm is used to periodically revise the forecasted load demand and the generating capacities to determine the electrical deficit.