The resolution of seismic data directly influences the characterization accuracy of oil reservoirs. To improve the resolution for effective sand body prediction, this study established a frequency enhancement technology process based on time-varying frequency-division deconvolution for thin oil-bearing sand bodies occurring in complex fault blocks. First, seismic signals were separated into different time windows, in which seismic wavelets were computed to obtain their amplitude spectra. Then, the corresponding seismic wavelets were deconvoluted within each time window to obtain the reflection coefficients. Finally, high-resolution broadband seismic signals were attained by integrating the reflection coefficients of the entire seismic data and convolving high-and low-frequency wavelets. This technology process was employed to process the actual 3D seismic data from the Wennan area of the Zhongyuan Oilfield. As indicated by the results, this technology process had a significantly elevated capacity to depict a single sand body by expanding the high-frequency effective information in acquired 3D post-stack seismic data, thus yielding high-quality data for the identification of thin sand bodies. Moreover, the prediction results were highly consistent with the actual drilling results. Therefore, the time-varying frequency-division deconvolution has great potential for application in complex fault blocks.