Abstract: The vacuum degassing process plays an important role in the production of high cleanliness steel, so it is extremely urgent to determine the different reaction sites of liquid steel under reduced pressure and how to reflect the overall degassing efficiency through reasonable parameters. Based on a similar kinetic mechanism, this paper experimentally simulated the vacuum degassing process of molten steel using the release process of dissolved oxygen (DO) in water. Under a vacuum pressure condition, a large number of small bubbles were observed to precipitate from the vessel’s internal wall or the surface of the oxygen probe. This phenomenon corresponds well to the internal degassing reaction assumption made in previous degassing mathematical models. To verify the existence of internal degassing sites, mechanical stirring was introduced to analyze and calculate the degassing rate at the bath surface and internal site. Results showed that the degassing rate at the bath surface is very low throughout the whole process and the bubbles that precipitated from internal degassing sites greatly improve the DO removal rate. Especially at a pressure of 25 kPa, the degassing rate is about ten times that at the bath surface. It was also confirmed that the internal degassing reaction mainly occurs in the initial stage of degassing, particularly in the range of high DO concentration. Moreover, the removal of DO is a first-order reaction process, and its volumetric mass transfer coefficient k · A · V^?1 is constant. Therefore, the removal process of DO can be used to simulate the degassing behavior of molten steel. To describe the effect of vacuum pressure and argon flow rate on k · A · V^?1, the correlation between log (k · A · V^?1) and log ε was determined by introducing the concept of stirring power density ε. Finally, the correlation was compared with the results from previous simulation studies.