Abstract: The effects of the initial molten steel level in tundish, filling flowrate through long nozzle and flow control devices on removing inclusions from continuous casting tundish during ladle changing were investigated via physical model testing with emulsion drops used to simulate inclusions. The results showed that the inclusions removed quickly in the range from 1/4 to 3/4 of the full depth which is calculated from the initial molten steel level in tundish, and the level of 3/4 of full depth seems to be an ideal working level for ladle changing. The inclusion removal rate increases with the increasing of the filling flowrate through long nozzle, and the removal of inclusions is improved significantly if the filling flowrate is higher than 2.5 m³/h. The removal of inclusions is further improved in tundish with a flow control device containing dam/wall especially a turbulence inhibitor used.