The integration of distributed new energy changes the power flow and increases system uncertainty in distribution grid. Aiming at the reliability evaluation problem of distribution grid with high penetration rate of new energy, an improved sequential Monte Carlo simulation method is explored to evaluate its reliability. First, the probability distribution models for wind and photovoltaic power generation are discussed. Then the adaptive importance sampling method is used in the Monte Carlo simulation process in order to reduce the sample variance and improve the computation efficiency of reliability indicators. Simulation experiments show that the proposed method can improve the computational accuracy and efficiency of reliability indicators, and high penetration rate of new energy can improve the operation reliability of distribution grid to some extent.