The increasing penetration of Plug-In-Electric Vehicles (PEV) in the existing electric distribution network needs strategic scheduling of their charging. On the other side, PEV's storage can be utilized for grid support through Vehicle-to-Grid (V2G) and Grid-to-Vehicle (G2V) functionalities. In this work, a multi-agent-based PEV control strategy is proposed for load flattening with consideration of voltage regulation and customer revenue. The power system is viewed as a multi-agent-based energy management system (MAEMS) consisting of three layers: Central Agent (CA), Aggregator Agent (AA) and Distributor Agent (DA). This work mainly focuses on only DA level where load flattening using PEV's storage is the main objective. Zones of energy need for load flattening are identified where charging and discharging of PEVs is required in order to achieve load flattening. In each zone, the total available energy from PEVs is distributed optimally among all the intervals using Water Filling Algorithm (WFA). During Optimal Enrgy Distribution (OED) using WFA, uncertainty of PEV availability for grid support has been taken into account for estimating the available PEVs energy capacity in each zone (charging and discharging). Multi-Objective Genetic Algorithm (MOGA) is used for setting optimal power transaction (OPT) between the grid and PEVs with two objectives: load flattening and voltage regulation. Pareto-front obtained from MOGA is used to decide OPT to achieve flat load profile by ensuring bus voltage limits. Further, an Adaptive Neuro-Fuzzy Inference System (ANFIS) based vehicle prioritization is implemented in order to maximize storage usage and to minimize cost of charging (CoC). The results show the impact of OED and OPT on load flattening and voltage regulation ahave been studied. Different cases are studied in ANFIS prioritization and the resultant effect on total PEV power availability and CoC are also studied. © 2019 Elsevier B.V.