Clustering the large scale active distribution networks (ADNs) into a set of smaller microgrids (MGs) or multi-microgrids (MMGs)-based operation of ADNs can have several benefits for the utility, electric power consumers, and distributed generation (DG) owners such as easier control strategy, distributed control among MGs, load routing and transfer among MGs and reliability enhancement. This research proposes a novel program for optimal operation of ADNs based on the MMGs approach. Initially, the operation of ADNs was carried out by integrated management of all distributed energy resources (DERs) using probabilistic forward-backward load flow using Monte Carlo simulation (MCS) algorithm. In this stage, energy storage devices (ESDs) as one of the significant components of MG are sized and sited in ADN. In the second stage, the operation quality is scrutinized accurately by testing the possibility of MG construction in the modified ADN based on various technical criteria such as adequacy, efficiency, voltage profile, and reliability. This issue has not been addressed in the previous researches. Finally, the results of the proposed model are compared with conventional operation method in different scenarios by implementation on IEEE 33-bus ADN and an actual Portuguese distribution network using a powerful multi-objective optimization tool known as non-dominated genetic algorithm-II (NSGA-II). (C) 2016 Elsevier Ltd. All rights reserved.