With　the　development　of　industrial　manufacturing,　flexible　manufacturing　system　provides　an　effective　way　to　improve　the　machining　quality　and　efficiency,　as　well　as　reduce　the　manufacturing　cost.　Fflexible　manufacturing　system　reliability　has　been　one　focus　of　recent　studies.　But　a　systematic,　effective　and　executable　reliability　evaluation　method　has　not　yet　proposed　in　existing　researches.　This　work　firstly　presented　a　three-layer　index　system　for　flexible　manufacturing　system　reliability　evaluation,　which　is　composed　of　the　whole　layer　indexes,　subsystem　layer　indexes　and　equipment　layer　indexes　determined　by　system　composition　and　running　features　analysis.　Time-varying　weights　of　each　layer　are　calculated　through　the　modelling　of　the　reliability　and　reliability　importance　based　on　the　classical　two-state　failure　hypothesis.　Then,　considering　multi-state　failure　modes　of　elements,　a　fuzzy　evaluation　model　of　flexible　manufacturing　system　reliability　is　established　based　on　fault　grading　and　statistics,　considering　grade　and　number　distributions　of　faults　within　a　certain　statistical　period.　Moreover,　through　a　simulation　case　of　a　box-parts　finishing　flexible　manufacturing　system,　the　influence　of　failure　rates　and　service　time　on　the　index　weights　distribution　is　firstly　revealed,　which　can　provide　theoretical　basis　for　the　improvement　of　flexible　manufacturing　system　reliability;　then　the　effectiveness　of　the　proposed　method　is　verified　through　the　comparison　of　results　obtained　by　the　previous　model　based　on　two-state　failure　and　the　proposed　model　based　on　multi-state　failure.