Photovoltaic（PV） generation is still considered to be an energy-replacement rather than capacity-replacement source because of its intermittent nature. Given that power system reliability can be elevated after integrating PV generation to the system， the PV generation should have a capacity credit to some extent. Evaluating this capacity credit is one of the urgent problems when planning for large-scare PV station integration into a traditional power system. A new method for evaluating the capacity credit of a PV generation station is proposed. Firstly， an annual PV generation fluctuating model is given. This model considers weather random conditions including the probability of different weathers， the maximum proportion of solar radiation under various weathers， the radiation fluctuation range due to clouds mask and the temperature coefficient random variation range. Secondly， a reliability index computation method based on the sequential Monte-Carlo simulation is proposed. The secant method is used as the iteration algorithm to get the capacity credit of the PV generation station， in which the convergence criterion facilitates testing two large samples means equal（the u testing） of hypothesis testing methods in the mathematical statistics. Finally， the proposed model and algorithm are validated with a benchmark reliability system.