This paper addresses a disassembly lot-sizing problem under stochastic batch demand for multi-products with a two-level structure. The proposed approach is based on a Batch Deterministic and Stochastic Petri Net (BDSPN) as a modeling tools to solve this type of problems under a stochastic process. The main goal of the problem is to determine which batch demand to be started disassembled first for each type of product to ensure the satisfaction of the stochastic batch demand of root-items over multi-criteria. An analytic method for the BDSPN is presented and applied to our model. After that, some performance indicators for each scenario are calculated. These indicators provide a range of measures to evaluate the performance of the proposed approach. After that, a new mathematical model is suggested to minimize the sum of extra costs, disassembly operations and holding costs for various products. A decision support tool, called PROMETHEE II method has been put forward to prioritize the different scenarios. To test the efficiency of the proposed model, a real case study of manufacturing industry related to CODIMATRA is simulated. The simulation results improve the capability of the suggested approach to satisfy the time-varying demands of leaf items over the planning horizon.