Monsoon onset information plays an important role in setting up planting strategy for achieving optimum yield. This study aimed to develop forecasting model for the monsoon onset in main rice growing area of Java used Regional Climate Model Version 3.1 (RegCM3). The forecasting models of the monsoon onset and September-Oktober-November (SON) rainfall data were developed using regression model that have the highest coefficient determination and the models were tested using likelihood ratio test. It was found that the forecasting models of the monsoon onset and September-Oktober-November rainfall data were polynomial orde 2 or cuadratic that have coefficient determination 69%, 74%, 80% and 86%. Likelihood ratio test found that RegCM3 rainfall data was not significantly different with observation rainfall data (α = 0.05). Onset in Java between 25th until 34th of 10-days period (early September until early December).