Abstract: Seismic wave CT inversion technology is an important method for rock burst hazard prediction, as well as routine rock burst prevention monitoring and effectiveness evaluation. However, in practical applications, active CT inversion has high implementation costs, cannot achieve continuous real-time monitoring, and is limited in large-scale detection, making it difficult to dynamically track stress changes. Passive CT inversion suffers from large errors in source location and low resolution, and the quality of tomographic imaging is constrained by the frequency and energy of natural microseismic events. To address this problem, a joint inversion strategy based on active-passive dual-source CT suitable for predicting rock burst hazards in coal pillars of deep roadways is proposed. First, active CT detection was implemented by arranging artificial seismic sources and receiver arrays, and a high-precision three-dimensional initial velocity model was obtained using the travel-time tomography method. Based on the active detection, the passive seismic sources recorded by the microseismic monitoring system during the subsequent time period, along with their ray travel-time information, were combined to form the joint inversion dataset. The high-precision three-dimensional initial velocity model obtained from active CT inversion was used as the initial model for the joint inversion. Travel-time tomography was performed again. The rock burst risk zones in the coal pillars were predicted according to the inversion results. An engineering application was carried out at the working face 7305 of Zhaolou Coal Mine, and the inversion results of active CT, passive CT, and active-passive dual-source joint CT were comparatively analyzed. The results showed that the dual-source CT joint inversion strategy effectively complemented the coverage blind zones of a single method and improved the identification accuracy of high-stress zones. Verification results based on microseismic events showed that more than 80% of the seismic sources were located within the high-velocity zones identified by the dual-source CT inversion, demonstrating the feasibility and accuracy of this strategy.