The present thesis was to study non-metallic inclusions during vacuum degassing in ladle treatment. Emphasize was mostly given to liquid CaO-Al2O3 inclusions. A series of industrial experiments were carried out at Uddeholm Tooling AB, Hagfors, Sweden. To gain an insight into the industrial findings, laboratory investigations were also performed. Large number of steel samples were collected and examined. Liquid calcium aluminate inclusions with low SiO2 and high SiO2 were often found with spinel inclusions before vacuum degassing. Laboratory experiments showed that spinel would react with the dissolved Ca in the liquid steel forming calcium aluminate inclusions. This laboratory results agreed with the industrial observation that spinel phase was quite often found in the center of the calcium aluminate phase. After vacuum degassing, most of the inclusions were calcium aluminate liquid inclusions. When dissolved Al level was low, 2 types of liquid calcium aluminate inclusions with considerably different SiO2 contents were found to coexist even at the end of the process. In view of the lack of the thermodynamic data for SiO2 activities in the low silica region, thermodynamic measurements were conducted in the CaO-Al2O3-SiO2-MgO system. The experimental results could reasonably explain the coexistence of the two types of the liquid oxide inclusions. While the total number of inclusions decreased during vacuum degassing, the number of bigger inclusions (>11.3 μm) increased generally in used ladles. This finding was in accordance with the previous studies, wherein, ladle glaze was found to be responsible for the supply of bigger inclusions. The behaviors of several types of inclusions in liquid steel were examined using a laser scanning confocal microscope (LSCM). While alumina inclusions tended to impact on each other, agglomerate and grow very quickly, none of the other types of inclusions, such as spinel and calcium aluminate was observed to agglomerate. The results of LSCM study agreed well with the industrial observation. Examination on a huge number of inclusions did not show any indication of impact and physical growth of the inclusions, except the alumina inclusions. The removal of inclusions around open-eye in a gas-stirred ladle was experimentally studied by a cold model with oil and water. Most of the simulated inclusions were brought up to the oil phase by gas-water plume. Inclusion removal into oil layer took place when the inclusions passed through the sphere-bed of the oil layer around the open-eye. A calculation showed that the contribution of metal-gas plume in inclusion removal was much larger than that of buoyancy mechanism. The results of the industrial experiments revealed that the inclusions distribution strongly depended on stirring condition. When a ladle was stirred by both gas and induction, inclusion removal near slag layer was significant. / QC 20100803