The study designed three different experiment scenarios to examine the braking behavior results under different traffic conditions, and made use of driving simulator to make experiment. The three traffic conditions included car following, crossing vehicles, and traffic signs condition. By collecting experiment data and statistics analysis, the study can find driving factors which real influence braking behavior under the three traffic conditions. Finally the study made use of Fuzzy theory and Back-Propagation Neural Network to set the driving behavior models. Comparing between Fuzzy inference model and BPN forecasting model by RMSE, the result showed no significant evidence to know which is better. However both of them had good forecasting ability.