Abstract: Natural microtremor wavefileds are dominated by ambient noise without earthquake events and natural microtremor technology by using surface or reflection waves extracted from ambient noise is a green and low-cost seismic exploration method to effectively image sub-surface structure and invert velocity information for near-surface and deep resources. Seismic reflection wavefields, which can obtain an exploration result with a higher resolution than surface waves, are difficult to be extracted with high fidelity from natural microtremor wavefields. To fulfill the high-precision natural microtremor technology with seismic reflections in sandstone-type uranium deposit, the key factors influencing the extraction quality of seismic reflections were analyzed by using simulated data in this paper. Firstly, a seismic velocity model with stratigraphic pinch-out for sandstone-type uranium deposit in Songliao Basin was built based on the result of seismic reflection exploration with active sources. Secondly, seismic acoustic wavefield simulation based on natural microtremor was achieved by using natural microtremor sources with random characteristics and finite-difference forward modeling technology, the simulated microtremor wavefields show good consistence with field microtremor wavefields. Thirdly, the basic principle of seismic interferometry method for retrieval of seismic reflections from microtremor wavefields was introduced, and it is shown that the noise sources located in the stationary-phase areas are constructive to retrieval of seismic reflections. Next, seismic reflection reconstruction from the above simulated microtremor seismic waves was achieved by using seismic interferometry, and the characteristics of retrieved seismic reflections in sandstone-type uranium deposit were analyzed. The qualities of seismic reflections reconstructed from simulated microtremor data with different recording length show that a longer recording time is better for the retrieval of seismic reflections from natural microtremor wavefields, and only shallow and near-offset seismic reflections are clearly retrieved from microtremor wavefields, which induces the good imaging result of stratigraphic pinchout but no reflection imaging of deeper targets. Then, seismic reflections with a higher quality were reconstructed based on a velocity model with increased velocity differences. The results indicate that the reconstructed seismic reflection supplies a better imaging result not only for shallow pinch-out but also for deeper targets. The above results of numerical simulation show that the length of data recording time and seismic wave velocity differences are the two key factors for the retrieval of seismic reflections, the natural microtremor data with a long recording time and formation media with high velocity difference can improve the quality of seismic reflection retrieval and imaging, which will support the high-precision natural microtremor exploration with seismic reflections in sandstone-type uranium deposit.