As the electricity industry has changed and became more competitive, the electricity price forecasting has become more important. Investors need to estimate future prices in order to take proper strategy to maintain their market share and to maximize their profits. In the economic paradigm, this goal is pursued using econometric models. The validity of these models is judged by their forecasting errors. This paper is an effort to compare the forecasting power of Artificial Neural Network (ANN), Genetic Algorithm (GA) and ARIMA models for hourly electricity prices in Iran electricity market. According to the results, ANNs has the best forecasting performance followed by GA in the second place and ARIMA model in the third place.