It is well established that nozzle blockage in Al-killed continuously cast steels is caused by the deposition of preexisting Al 0 inclusions at the narrowest part of the nozzle. Thisa adhere strongly at the wall by interfacial forces. The theory examined in this thesis is that during the curved flow path through the nozzle, inclusions move towards the nozzle wall by the action of centripetal force. A mathematical model has been developed which takes account of centripetal, gravitational, and drag forces. A computer programme has been written to predict particle behaviour in nozzle flows. To study the behaviour of second phase particles, a water model using a perspex tundish and an interchangeable glass nozzle has been built. Turbulence has been almost completely removed from the experimental tundish. Hydrogen bubbles electrolytically generated in water have been used to simulate the inclusions in steel and their flow paths through the nozzle have been recorded by cine photography. With frame by frame analysis of the films, the effects of bubble size, (50 pm to 1000 um), and flow velocity have been studied. Two experimental nozzles have been tested which have demonstrated the importance of limiting the centripetal force. It has been shown that there is a good relationship between hydrogen bubble behaviour and model predictions. This has been particularly true of small bubbles near the nozzle wall. By comparing the results with other experimental work it has been shown that centripetal forces can have a significant effect on the rate of blocking. Practical suggestions for reducing the turbulent behaviour in steel tundishes and nozzles are made. The merits of an improved nozzle design which takes advantage of the results of the model work are discussed.