The proposed intelligent 7 days natual gas load forecasting system for a power supplier of a city is based on a combination of artifical neural network technique and correlation analysis logic. The forecasting system is composed of two netual networks, each representing a period of a week. Each network can be sensibilized using different input Information for each period of a week. The input data is determined by degree of the correlation coefficient The proposed revised neural network method shows more improved prediction results than by traditional methods.