<p>Sensor-based sorting is one of the effective methods to solve the high dilution rate of ores. However, gold ores generally have problems such as low grade and fine particle size of gold, which make it difficult for detectors to directly detect gold and limit the application. This study takes a gold deposit in Shanxi as an example, demonstrating a method for gold ore sorting based on process mineralogy and gold-hosting minerals. Through the analysis of process mineralogy, the types and occurrence characteristics of gold-hosting minerals in the gold ore and their correlation with the Au grade were investigated, and the potential for the detection of gold based on gold-hosting minerals was discussed. Subsequently, XRT response characteristic tests were conducted to analyze the differences in XRT images of different ore particles and evaluate the sortability. Finally, the feasibility of XRT sorting for gold ores was verified by a sorting experiment. The results show that XRT sorting can achieve good performance indicators, such as a discard rate of 40.53% and an Au loss rate of 5.46%. The identification and sorting of gold ores by detecting gold-hosting minerals have significant effectiveness, which depends on the good correlation between the Au grade and the grade of gold-hosting minerals, as well as the XRT response of different ore particles.</p>