Abstract: Due to the characteristics of nature, wind power output is often negatively correlated with load demand. So, in North China with rich wind resources, the wind power curtailment problem has already emerged and become more and more serious due to lack of deep peak-shaving and start-stop peak-shaving units.The peak-shaving requirement has become one of the focused factors to evaluate the wind power accommodation capacity of power system. In the planning stage, the worst case always tends to be considered, which assumes that it has no wind power output during peak load period in the daytime and rating output during valley load period in the night. This practice actually amplifies the negative effects of wind power and limits the development of wind resource. Therefore, the uncertain nature of wind power generation as well as conventional generators and loads should be considered and the probabilistic analysis methods should be used to analyse the peak-shaving requirement of large scale wind power integrated system. Peak-shaving problem of large-scale wind power integrated system, with more obvious timing sequence characteristics, is suitable to be analysed through sequential monte-carlo method which can effectively consider the factors related to timing sequence. In North China, the conventional power generators are still the main peak-shaving resources, which usually be simulated by two state model in monte-carlo simulation. This model can only consider the up state and forced outage state of generators, without regarding to the planned outage state. Therefore, after each state sampling of the generators, it is necessary to carry out the unit combination calculation, and causes the monte-carlo simulation being more time consuming and less practical. In this paper, a peak-shaving requirement analysis method based on sequential monte-carlo simulation for large scale wind power integrated system is presented after analysing the influence of large-scale wind power on the peak-shaving characteristics of power system. This method establishes the time sequence model of wind power and load, as well as the conventional generators of which the planned outage is also considered as a random variable, and a three-state transfer sequence model is established.Then the peak-shaving shortage indices of large scale wind power integrated system are defined, which can be used to probabilistically evaluate the peak-shaving requirement of the system. The convergence of the method is verified by IEEE-RTS79 system. At last, the method is applied in the real large power system of Jibei power grid in North China.