Train　regulation　plays　an　important　role　in　urban　metro　transportation,　and　most　existing　studies　on　train　regulation　are　based　on　centralized　control.　Motivated　by　the　real-time　control　demand　and　the　rapid　development　of　vehicle　based　train　control　(VBTC)　technology,　this　paper　investigates　the　train　regulation　problem　by　employing　distributed　model　predictive　control　(DMPC).　We　firstly　present　a　distributed　control　framework　for　train　regulation　in　metro　loop　lines,　where　each　train　is　assumed　self-organized　with　the　capability　of　computation　and　communication　with　its　predecessor.　Then　we　propose　a　DMPC　algorithm　for　train　regulation　in　metro　loop　lines,　in　which　each　train　decides　its　control　input　by　optimizing　a　local　cost　function　subject　to　operational　constraints.　We　finally　provide　numerical　examples　to　verify　the　effectiveness　of　the　proposed　DMPC　method,　showing　that　it　exhibits　comparable　performance　with　the　centralized　MPC　and　its　computation　cost　is　significantly　reduced.