Regional　vulnerability　research　is　helpful　in　understanding　the　formation　mechanism　of　regional　disaster　risk　deeply.　Owing　to　the　pressure　from　earthquake　disasters　occurred　frequently　in　the　whole　world,　the　assessment　of　regional　vulnerability　to　seismic　hazards　is　a　focus　in　current　research　of　seismic　disaster　risk.　This　paper　presents　a　nonlinear　dynamic　model　for　assessing　regional　vulnerability　to　seismic　hazards　based　on　maximum　flux　principle　(MFP).　Firstly,　a　vulnerability　index　system　was　designed　for　the　urban　agglomeration　of　Beijing,　Tianjin　and　Hebei　in　China,　and　the　index　data　were　obtained　based　on　the　Statistical　Yearbook　and　other　materials.　The　vulnerability　indexes　constituted　a　complex　system,　and　the　indexes　were　taken　as　the　components　of　the　complex　system.　Secondly,　based　on　the　information　theory　and　MFP,　a　dynamic　equation　that　controls　the　generation　and　evolution　of　the　complex　system　was　derived　theoretically.　The　regional　vulnerability　evaluation　was　realized　by　conferring　reasonable　weight　to　the　indexes　to　make　the　system　become　stable.　Finally,　the　self-organizing　neural　network　was　used　to　simulate　the　mechanism　model　to　implement　vulnerability　assessment.　The　assessment　results　showed　that　Langfang　and　Chengde　were　at　high　vulnerability　level;　Qinghuangdao,　Cangzhou　and　Zhangjiakou　were　at　moderate　vulnerability　level;　Baoding,　Shijiazhuang,　Beijing　and　Tangshan　were　at　lower　vulnerability　level;　Tianjin　was　at　low　vulnerability　level.　It　reveals　that　the　investment　in　disaster　prevention　was　in　an　unbalanced　situation　among　all　the　cities　in　Beijing-Tianjin-Hebei　metropolitan　area.