High-order fuzzy time series is a popular method to forecasting, but it has lacked convenient tools for researchers to simulate their design models. This paper is presented a new technique to develop the simulation tools on Simulink using customs s-function. This technique can separate to three modules that are fuzzifier module, forecasting module and defuzzifier module to increase the flexibility. For accuracy, this simulation has applied to forecast the enrollments of the University of Alabama and foreign exchange rate of EURO to US$. The results show number of order and length of learning time series have effected to forecasting error when number of order is little than length of learning time series.