Abstract This paper proposes a day-ahead self-healing scheduling approach in isolated networked microgrid (NMG) systems. The proposed approach is based on a two-level flexible energy management system (EMS). The upper-level EMS is responsible for optimal scheduling of the normal-operated MGs, while the lower-level help the MGs for operating on-fault in self-healing and islanded modes. When a fault occurs in an MG, it divides the on-fault MG into two zones. The grid-connected zone receives power support from the other normal-operated MGs. Furthermore, a secure optimization problem is presented to operate the separated zone in an islanded operation. The proposed approach is implemented on a test system with five NMGs. To cope with the operational uncertainties, a stochastic programming approach is applied. The simulation is run over a 24-hour scheduling time horizon. The results of the simulations in the normal, self-healing, and islanded operation modes for normal-operated and on-faulted MGs in cases without and with DRP are presented. The results reveal the effectiveness of the proposed stochastic EMS model in enhancing the performance of the network and decreasing the operational costs of the MGs.