The　hybrid　approach　of　CREAM　(cognitive　reliability　and　error　analysis　method),　fuzzy　theory　and　Bayesian　network　(BN)　was　constructed　to　analyze　the　reliability　of　scaffolding　operators　in　this　work.　A　construction　site　in　Fuzhou　City　(Fujian　Province,　China)　was　targeted　as　an　example.　Firstly,　seven　operational　and　five　cognitive　subtasks　were　developed　by　the　characteristics　of　scaffolding　operations.　Then,　the　relational　network　of　the　common　performance　conditions　(CPCs)　was　determined　by　a　BN.　Expert　evaluation　can　obtain　a　priori　prob-abilities.　Fuzzy　languages　were　transformed　into　conditional　probabilities　by　fuzzy　theory.　The　probability　distribution　of　the　control　model　can　be　determined　based　on　the　relational　network.　Secondly,　the　human　error　probabilities　(HEPs)　of　the　operational　subtasks　were　calculated　based　on　the　center　of　the　area　(COA).　Based　on　the　influence　of　the　operational　subtask　on　the　cognitive　subtask,　the　HEPs　of　the　cognitive　subtasks　were　calculated　with　BNs.　Finally,　the　total　HEP　of　the　scaffolding　operation　was　identified　using　a　series-parallel　hybrid　structure.　The　results　show　that　the　total　HEP　of　the　scaffolding　operation　at　this　construction　site　is　4.30E-03.　The　scaffolding　dismantling　and　supervisory　inspection　have　a　significant　impact　on　human　error.　Also,　site　management　should　strengthen　training　and　education　on　work　methods.　This　method　can　provide　a　new　idea　for　human　reliability　analysis,　as　well　as　a　new　perspective　for　the　safety　of　work　at　height.