In assembly line balancing problems, parallel execution of assembly operations is often advocated because of its enhanced flexibility and minimum lead-time. Although the theoretical maximum number of possible assembly sequences combinatorially explodes with the number of components in a product, graphical representations can depict these sequences in a surveyable way. The AND/OR graph representation is an appropriate basis for optimum sequence selection, which can be achieved via heuristic, metaheuristic, and exact methods. The exact method, based on binary linear programming, is described. To arrive at the appropriate model, a novel approach for AND/OR graph generation, based on subassembly detection, is presented. The method is demonstrated with simple cases and next extended to increasingly complex products. A modification of the optimization method is applied, which enables a search for sequences with maximum parallelism.