Flexible　disassembly　line　design　for　end-of-life　(EOL)　products　is　a　key　issue　in　the　remanufacturing　industry.　However,　existing　studies　for　disassembly　line　balancing　have　not　simultaneously　considered　multiple　EOL　products,　the　identical　parts　of　these　products　and　uncertainty　during　disassembly,　which　are　important　characteristics　of　flexible　disassembly　lines.　The　present　study　addresses　a　new　flexible　multi-product　disassembly　line　balancing　problem　in　which　(1)　disassembly　schemes　need　to　be　selected,　(2)　a　workstation　can　disassemble　multiple　EOL　products,　(3)　identical　parts　of　multiple　products　can　be　treated　as　identical　tasks,　and　(4)　only　partial　probability　distribution　information　of　processing　times　is　known.　For　the　problem,　an　integrated　approach　is　developed,　which　is　composed　of　a　chance-constrained　program,　a　distribution-free　model,　efficient　valid　inequalities　and　an　exact　lifted　cut-and-solve　method.　Numerical　experiments　are　conducted　on　an　illustrative　example,　10　instances　based　on　realistic　products　and　480　randomly　generated　instances　with　up　to　20　products,　400　tasks　and　86　workstations.　Computational　results　show　that　the　proposed　valid　inequalities　can　reduce　about　75%　computational　time　of　the　original　model,　and　the　lifted　cut-and-solve　method　needs　only　17.53%　and　40.65%　of　the　computational　times　required　by　the　CPLEX　and　the　classic　cut-and-solve　method,　respectively.