Intelligent methods have been developing for several decades, which through their imitation of natural processes and human intelligence, realize numerous advances in many different areas of application. From medical diagnostics, education, health, entertainment to autonomous control of vehicles and robotics. Methods od artificial intelligence (intelligent methods) are also used in economics in form of business intelligence – utilizing a large amount of company's data that results in better business decisions, and thereby increasing business success. There are also different fields of application of intelligent methods in economics, eg. in stock price movements, risk forecasting, segmentation of customers, targeted advertising, consumption forecasts and so on. Artificial Neural Networks belong to intelligent data mining methods that have proved useful in predicting economic indicators as well as in decision support for middle and higher management levels in making business decisions. The purpose of the paper is to develop and compare two models of artificial neural network based on available data on average natural gas consumption of public buildings in the Republic of Croatia over the last three years, with the aim of forecasting natural gas consumption and revealing important predictors of gas consumption. The theoretical part of the paper defines the difference between biological and artificial neural networks, gives characteristics and describes architecture of neural network models. In the practical part, Statistica is the tool used for analysis and data processing, apropos building neural network models.