To assess the impact of large-scale new energy sources on the probabilistic power flow (PLF) of power systems, a Markov Chain Monte Carlo (MCMC) PLF method based on Copula theory, slice sampling and Latin hypercube sampling is proposed. The probabilistic model of the correlative input variables is established by the Copula theory with the Kendall rank correlation coefficient used to measure the correlations. The sample space of random input variables is obtained by slice sampling and the Latin hypercube sampling is further introduced to deal with the initial samples to improve efficiency. The modified IEEE 14-bus system is used as an example to demonstrate the correctness and effectiveness of the presented method and the influence of correlations between wind and photovoltaic power outputs on PLF is studied. The results show that wind and photovoltaic combined increases the reliability and economy of system operation and consideration of the correlations will provide a more accurate assessment on the effect of wind and photovoltaic outputs on PLF.