Increasing penetration of EV integration to the distribution side of power systems is dramatically reshaping the distribution grids. Not only are EV chargings comparatively large loads; but also, with the concept of Vehicle-to-Grid (V2G), EVs can feed back the power stored in the battery to the grids, switching EVs from traditional loads to distributed generation. Compared with stationary storage, the primary challenge behind utilizing the energy stored in EVs lies in the uncertainties that vary for different owners and at different times: the arrival time, departure time and energy demands (for charging). This chapter proposes a Demand Side Management (DSM) strategy under which a Distribution Grid Operator (DSO) would target an optimal utilization of the energy stored in the EVs. The stochastic modeling of the uncertainties in EVs is developed in the context of a distribution grid. A Monte Carlo simulation method is applied to make the problem tractable. Furthermore, by tightly relaxing the boundaries of the constraints, we further convert the original non-convex problem into its convex counterpart. Finally, the methodology is tested using real-life data collected on the UCLA campus. Through extensive analysis we examine the merit of the proposed stochastic model, and the effect that the reverse power flow of EVs has on the distribution grids.