To improve the time-domain imaging of poor-quality seismic data, the common-reflection-surface (CRS) stack method was introduced to simulate zero-offset (ZO) stacked sections from a multicoverage data set based on automatic coherence analysis of seismic signals. This method produces improved ZO stacked sections with a high signal-to-noise ratio (S/N) and good continuity of reflection events. However, the stacking results may have some undesirable artifacts that can degrade the poststack migrated image. To overcome these drawbacks, I have developed a prestack data regularization method, based on CRS partial stacks, which produces prestack data with high S/N and enhanced reflection events. The regularized data are usually applied for velocity analysis and conventional prestack migration in the time and depth domains. Recently, the CRS stacking operator has also been applied for developing a new type of prestack beam migration. This new migration combines the classic Kirchhoff migration with the CRS stack method, in which the beam-forming process stacks locally coherent events that are performed using the CRS operator during migration. This work reviews this CRS-based prestack migration method in the time domain and presents a comparative study with the main standard applications of the CRS stack method, such as CRS stacking plus poststack time migration and CRS-based regularization plus prestack time migration (PSTM). To evaluate its effectiveness and reliability, CRS-based PSTM and CRS-based prestack data regularization were applied in a crooked line. The time-migrated image resulting from the regularized data has strong migration artifacts due to the crookedness of the seismic line; in contrast, the CRS-based time migration provides a good-quality image without migration artifacts.