Abstract: In this paper, the GC-MS ion selection method was used to analyze the diagnosis ratio parameters of six different oil spill samples. Six kinds of oil spill samples were identified by multidimensional scaling method, and compared with the repeated limit method. Results show:Multidimensional scaling method can be used to effectively carry out a large number of samples of oil fingerprint analysis at the same time, it can directly show the relationship between the object of study, and realize the classification and identification of oil samples. Compared with the repeatability limit method, the method can simplify the tedious calculation process of comparison between two samples and avoid miscarriage of justice in the case of multiple diagnostic ratios of different oils, thus effectively to make up for the deficiency of the method of repeatability limit and other methods. It can provide some references for the marine oil spill accident fingerprint identification.