Abstract: In view of the great harm to the human body of pesticides, the national food safety standards and the International Food Code Committee standards limit the maximum residue limits of pesticides in soybean oil. Gas chromatography-mass spectrometry (GC/MS) is often used for the identification and quantitative detection of the compound, because of the characteristics of high separation efficiency. However, the current GC/MS is mainly used in the combination of single quadrupole GC/MS and selective ion monitoring (SIM) technique. The selectivity of SIM acquisition mode is poor, and it is easily disturbed by the matrix. The multiple reaction monitoring (MRM) technique of the triple quadrupole gas chromatography-mass spectrometry （GC-MS/MS） can effectively eliminate the interference of the matrix, which is beneficial to the determination of pesticide residues in complex matrix. A reliable pesticide residues analysis method in soybean oil was developed in this paper, based on on-line gel permeation chromatography coupled with triple quadrupole gas chromatography-mass spectrometry （GPC-GC-MS/MS）. GPC system served as on-line purifying pretreatment to remove most part of remained fat in soybean oil and effectively reduced the time of sample pretreatment. At the same time, multiple reaction monitoring (MRM) acquisition mode of GC-MS/MS could greatly reduce the matrix interference and provide much better selectivity. The results show that the relative coefficients are above 0.999 with the most of pesticides standard solutions from 1 to 50 μg/L. The method detection limits range from 0.02 μg/kg to 2.59 μg/kg, and the quantitative limits range from 0.08 μg/kg to 8.62 μg/kg. The recoveries are 70.4%-119% with added concentration as 10, 50, 300 μg/kg. By comparing the experimental results of GPC-GC-MS/MS, GPC-GC-MS and GC-MS/MS, the MRM data acquisition mode of GPC-GC-MS/MS can get the lowest detection limit and the best recovery of pesticide residues in soybean oil. The detection method based on on-line GPC and GC-MS/MS can simplify the process of sample pretreatment, reduce the matrix interference, improve the selectivity and sensitivity of the analysis, and which can be applied to the qualitative and quantitative analysis of trace residues in complex matrix.