An　assembly　line　is　an　industrial　arrangement　of　machines,　equipment,　and　operators　for　continuous　flow　of　workpieces　in　mass-production　operations.　In　an　assembly　line　balancing　problem,　tasks　are　allocated　to　workstations　according　to　their　processing　times　and　precedence　relationships　amongst　tasks.　Nowadays,　some　research　investigated　the　reliability　of　assembly　production　by　taking　account　of　task　time　uncertainties.　Our　research　utilizes　uncertainty　theory　to　model　task　time　uncertainties　and　introduces　the　belief　reliability　measure　to　the　assembly　line　production　for　the　first　time.　We　proposed　a　multi-objective　optimization　model　that　aimed　at　maximizing　the　belief　reliability　and　minimizing　the　cycle　time.　The　problem　is　solved　using　a　newly　developed　restart　neighborhood　search　method.　The　numerical　experiments　are　conducted　to　verify　its　efficiency.　The　methodology　proposed　in　this　paper　is　applicable　to　any　industry　(including　the　automotive　industry)　when　the　historical　data　on　task　processing　times　are　very　scarce.