Oil tankers carry large quantities of liquefied chemical cargoes that are flammable, explosive and/or toxic. Hence, a collision with a tanker that causes an oil spill poses a severe threat to the marine environment and human life. In order to quantify and analyze the risk factors of ship collision oil spill, this paper adopts a combination of knowledge element (KE) and dynamic Bayesian networks (DBN) to conduct an emergency scenario study based on the “scenario-response” model. Firstly, the key elements of “accident scenario state, human factors, emergency measures, and emergency goals” are selected to represent the accident. Then, the mechanism of accident evolution is analyzed according to the case, and DBN is used to build a scenario model of oil spills from tanker collisions. Finally, to verify the importance of human factors and the scientificity of emergency measures, the oil spill accident due to the collision between the two vessels known as MT “SANCHI” and MV “CF CRYSTAL” is used as an example for analysis. The accident model deduction results are in line with reality, and the research results help relevant decision makers to understand the deduction process of oil spills from tanker collisions, which is of great significance to enhance the safety of oil tanker shipping and marine environmental protection.