Microgrid is a controllable system that supplies power generated from some distributed units operating in an islanded or grid-connected mode to a community load. Renewable energy sources are beneficial to microgrids for clean energy generation. This paper proposes a memory-based multi-verse optimization (MMVO) algorithm for optimal power scheduling in microgrids among various generation units under load uncertainty. The random nature of the uncertain load is modeled using stochastic programming, and the improved algorithm is tested on two different scale microgrids, i.e., IEEE 37 -node and IEEE 141-node test systems. The simulation results demonstrate that the proposed algorithm minimizes the energy production cost under uncertain load demand. © 2023 IEEE.