Electricity price fluctuates wildly which makes huge risk. Accurately predicting the electricity price will contribute to market participants to manage risk and maximize the profit of participants. This paper applies ARMA—GARCH models to forecast day-ahead electricity prices in PJM electricity market and Nord Pool. With the assumption that residuals obey normal and student’s t distributions, this paper compares the prediction accuracy of different ARMA-GARCH models, and finds that in most cases the asymmetric GARCH models perform well. But the ARMA-GARCH models can not be applied in the electricity market in which price volatility is extremely huge and the correlation between electricity price serials is weak, such as Australia electricity market.