The aim of the paper is to propose an approach for statistical assessment of the potential of plug-in electric vehicles (PEV) for vehicle-to-grid (V2G) ancillary services, where it focuses on PEVs doing daily home-work commuting. In this approach, the possible ancillary services (A/S) for each PEV fleet in terms of its available V2G power (AVP) and flexible intervals are identified. The flexible interval is calculated using a powerful stochastic global optimization technique so-called ”Free Pattern Search” (FPS). A probabilistic method is also proposed to quantify the impacts of PEV’s availability uncertainty using the Gaussian mixture model (GMM), and interdependency of stochastic variables on AVP of each fleet thanks to a multivariate modeling with Copula function. Each fleet is analyzed based on its aggregated PEV numbers at different level of distribution grid, in order to satisfy the ancillary services localization limitation. A case study using the proposed approach evaluates the real potential in Niort, a city in west of France. In fact, by using the proposed approach an aggregator can analyze the V2G potential of PEVs under its contract.21 Keywords: Vehicle-to-grid, Ancillary services, Distribution grid, Gaussian Mixture model, Copula function, Free Pattern Search