Correct prediction of natural consumption is crucial for both gas distributors and consumers. By means of an accurate prediction, potential issues regarding natural gas distribution systems may be reduced, distribution limits may be planned properly. In this paper, an ampiric relation has been developed dependent on basic consumption indicators and meteorological data by making use of Artificial Neural Networks (ANN) which is used broadly in various disciplines and providing good prediction results in nonlinear multivariable models. With the proposed system, monthly natural gas consumption prediction study is implemented for Siirt which is one of the first-time users of natural gas in dwelling houses (share in total consumption 1.8 %, 2.5 % in terms of customer number). R2=1 statistical error value is found for analytical expression of provided prediction model.