Trace oxygen has a noticeable influence on the purity assessment of high purity gold when the total elemental impurities deduction method was used to calculate the purity. However, the previous elemental impurities deduction method did not calculate the non-metallic elements such as oxygen, making the purity assessment not persuasive. The inert gas fusion infrared absorption method was established to measure the content of trace oxygen in high purity gold reference materials. The secondary ion mass spectrometry was used to compare the methods to ensure the reliability of the measurement results. The measurement parameters of the ONH analyzer were optimized, and the optimal working conditions were confirmed as follows: purge time 35 s, analysis delay 75 s, exhaust cycle 2, exhaust time 25 s, exhaust power 4 500 W, analysis power 4 000 W. Tin was selected as the flux, and the ratio of gold to tin was determined to be 5∶3 by oxygen release experiment. The secondary measurement of the gold sample showed that the residual oxygen was consistent with the blank, indicating that the addition of tin particles could promote the release of oxygen in gold, thus solving the problem of incomplete release of oxygen in gold. The tiny particles were deoxidized repeatedly to reduce the blank, and a stable blank was obtained. The limit of quantitation of the method reached 0.1 mg·kg^-1. The calibration coefficient is 1.012, and the recoveries of oxygen are between 95% and 105%, which verifies the reliability of the measurement method and ensures the traceability of the measurement results. In the secondary ion mass spectrometer, Cs⁺ is used as the primary ion source, the aperture is 400 μm, the ion beam intensity is 3 nA, the beam spot size is about 20 μm, the grid scanning size is 10 μm, and the secondary ion aperture is 400 μm. After sputtering and ionization, ¹⁶O^- and ¹⁸O^- ion currents were collected. SRM685 high purity gold reference material was used as the measurement standard. The oxygen content was calculated through the cyclic measurement of the standard and sample by comparing the ion current intensities between standard and sample. The results of the two methods were (1.1±0.3) and (0.9±0.3) mg·kg^-1 respectively. The uncertainty evaluation showed that the primary sources were the certified reference materials and measurement repeatability. The two results were consistent within the uncertainty range. Finally, the trace oxygen content in the high-purity gold reference material was (1.0±0.4) mg·kg^-1. The accurate determination of trace oxygen in high purity gold was realized by the two established measurement methods, which provide effective methods for the determination of trace oxygen and the development of high purity gold and other high purity metal certified reference materials.