The rapid development of plug-in hybrid electric vehicles（PHEVs） and renewable generation brings new challenges to the secure and economic operation of power systems. A stochastic optimization based dispatching model， capable of accommodating uncertain outputs of PHEVs and renewable generation， is developed. The probability distributions of energy consumption and charging/discharging behaviors of PHEVs are first studied. The probability distributions of the wind and photovoltaic generation outputs are derived assuming that the wind speed follows the Rayleigh distribution and solar irradiance follows the Beta distribution. The mathematical expectations， second order expectations and variances of the power outputs of wind and photovoltaic generation are derived analytically. On this basis， a stochastic optimization dispatching model with the objective of reducing the fluctuations of renewable generation outputs is finally established. The well-established cross-entropy method is employed to solve this optimization problem. The feasibility and efficiency of the dispatching model and the cross-entropy method are demonstrated by a 33-bus distribution system.