Abstract: Alone with the participating of incentive demand response into power grid operation gradually, it plays significant role in improving load distribution and takes an important part in raising the reliability of distribution network. Based on the features of existing incentive demand response and according to the principles of consumer psychology, the non-schedulable response model for interruptible load and emergency demand response, and the schedulable model for direct load control, were respectively established. On the basis of different time scales and by means of fuzzy C-means clustering algorithm the incentive demand response resources were classified, and on this basis the response features such as the response capabilities and uncertainties of non-schedulable and schedulable resources under different time scales were considered, and then a flexibly configurable two stage response scheduling model with the participation of incentive demand response was established. At last, the reliability calculation for an improved RBTS system was performed by use of sequential Monte Carlo simulation. The effectiveness of the proposed method is verified by simulation results.