In the conventional dynamic economic dispatch methods， the spinning reserve allocation is based on deterministic criterion. It is hard to maintain response risk at a given level for all periods. A novel method is proposed to overcome the problem. The system spinning reserve requirement is no longer fixed but flexible in the new method， and it is adjusted with the system operating situation to achieve equal system response risk during dispatching time intervals. Expected demand not supplied（EDNS） is used as a response risk index. To ensure that the limit of EDNS be treated as one of constraints in the dynamic economic dispatching， a new expression of EDNS which includes｛0，1｝ variables is introduced. As such， the response risk evaluation can be embedded into the dynamic economic dispatching， and they can be calculated together. The model is a mixed integer programming problem. To solve the problem， additional constraints for binary variables are used to transform the mixed integer programming problem to a quadratic optimization problem. And decoupled primal-dual interior point method is used to achieve solutions. Test results show the efficiency of the proposed method.