This paper proposes a two-stage stochastic bilevel model for energy management of an active distribution network with multi-microgrids. The distribution network operator connected to the upstream high voltage grid and interacting with multi-microgrids has the task of scheduling the whole system to feed loads, providing safe operation and dealing with uncertainty due to changes in the loads and renewable generation. The multi-microgrids connected to the nodes of the distribution system aim to schedule their productions to feed their internal loads and exchange power with the distribution network operator in both day-ahead and real-time frames. With this purpose, a bilevel model is presented for modeling the interaction between the distribution network operator and the multi-microgrids in which the distribution network operator is located at the upper level and all multi-microgrids at the lower level. The proposed bilevel model is formulated as a mathematical program with equilibrium constraints that is transformed into a mixed-integer linear model through Karush–Kuhn–Tucker conditions and linearization rules. A case study illustrates the effectiveness of the proposed methodology.