The conventional deterministic methods can not accurately assess the active power output of the wind farm due to the randomness and intermittence of wind power, and the probabilistic methods are commonly used to solve this problem. In this paper the multi-state fault model is built considering run, outage and derating state of wind turbine, and then based on the turbine failure model, the reliability model of the wind farm is established considering the randomness of the wind speed and the wind farm wake effects. The active wind farm output probability assessment methods and processes are given based on the Monte Carlo method based on the above-mentioned model. The related programs are written in MATLAB, and the probability assessment for active power output of a wind farm is carried out, based on which the effects of reliability parameters and model parameters on assessment results are analyzed. The results show that the shape parameter and scale parameter of wind speed model have a greater impact on the results of the assessment.