Modern green products must be easy to disassemble. Selective disassembly is used to access and remove specific product components for reuse, recycling, or remanufacturing. Early related studies developed various heuristic or graph-based approaches for single-target selective disassembly. More recent research has progressed from single-target to multiple-target disassembly, but disassembly model complexity and multiple constraints, such as fastener constraints and disassembly directions, still have not been considered thoroughly. In this study, a new graph-based method using disassembly sequence structure graphs (DSSGs) was developed for multiple-target selective disassembly sequence planning. The DSSGs are built using expert rules, which eliminate unrealistic solutions and minimize graph size, which reduces searching time. Two or more DSSGs are combined into one DSSG for accessing and removing multiple target components. In addition, a genetic algorithm is used to decode graphical DSSG information into disassembly sequences and optimize the results. Using a GA to optimize results also reduces searching time and improves overall performance, with respect to finding global optimal solutions. Case studies show that the developed method can efficiently find realistic near-optimal multiple-target selective disassembly sequences for complex products.