This paper deals with the problem of disassembly line balancing where partial disassembly and uncertainty of task times are studied. Few papers have addressed the stochastic disassembly line balancing problem and most of existing work focused on complete disassembly and have not considered AND/OR graphs. In the present work, tasks of the best selected disassembly alternative are to be assigned to a fixed number of workstations while respecting precedence and cycle time constraints. Task times are assumed to be random variables with known probability distributions. An AND/OR graph is used to model the disassembly alternatives and the precedence relationships among tasks and subassemblies. The objective is to balance workstations’ idle times, i.e. differences among stations’ loads are as small as possible. A stochastic binary program is developed. To illustrate the applicability of the solution method proposed, it was performed on on a set of problem instances from the literature.