This study search for proper models to forecast Jakarta Composite Index (JCI) and then compare their forecasts. The stock index from strong markers, like Dow Jonc Industrial Average (DJIA) and NJKKEI, as well as the index from regional markets, like SET, are expected to have strong influences on JCI. More specijìcally, it is expected that SET will be able to explain the realocalion of short term fund from Thailand to Indonesia through capital market due lo unfavour political situation in Thailand. Other than thai, exchange rate ¡s also expected to have effect on JCI movements. By using the daily darafrom January 3, 2005 to January 2, 2006, the study found that the proper models to be used toforecast JCI are GARCH (2,2) Model and ARIMA (1, 1,0) Model. The empirical results showed thai the forecast from ARIMA Model is superior to that of GARCH Model.