Inflation is a condition where the level of price increases in general and continuously. To control inflation , the government through various policies.This study found that the pattern of the data plot is relatively constant and the deviation data is not constant.Mean models created namely; Inflation = 5.758472 + et. From the examination of the lag 1 residual plot can be seen that the p-value is equal to 0.0000, with an error rate of 5%, where research is a significant heteroscedasticity. After examination of up to lag 12 it produces a fairly small probability value of less than 5 %. Small probability value to a fairly long lag is indicative GARCH model is more suitable than the ARCH. Based on GARCH overfitting process it can be concluded orders p and q are used is p = 1 and q = 1. So that the mean and variance models are models created; Inflation = 5.386303 + et. Sehingga σ2t = 1.209697 + 0.810531e2t-1 – 0.322105 σ2t-2. Based on the results forecast for the next 5 years , it was found that the inflation rate will continue to fluctuate, The highest value reached 9.27% and the lowest value of 0.66 %