In order to statistically predict die swell, the study used the Monsanto processability test data which have monitored in 1993 and 1995. Each data was characterized by die swell characteristics obtained at each different shear rate in 4 zones. To select proper variables and to determine die swell univariate statistical analyses were studied in the beginning, and afterwards various applied statistical techniques such as partial correlation, factor analysis and regression analysis have been studied. To obtain model equations of 2 dependent variables (running and relaxed die swell) with 71 independent variables (raw materials and test conditions) independent variables for multicollinearity were simplified and multi-regreszion and residual analysis were repeatedly applied in this study. Simulation system was built to predict die swell properties of unvulcanized rubber depending on the design of materials' ingredients and process factors by using the statistical methods before actual test. Simulation system was composed of forecasting result, trend analysis and result analysis with graphic user interface (GUI).