Stacking may be considered one of the most important steps in seismic processing providing an approximate zero-offset section with an enhanced signal-to-noise ratio in which the NMO velocity is the key quantity. In this work, we will compare the conventional and automatic approaches to NMO velocity estimation employing both CMP and CRS type of stacking of a 2D real marine data set. Moreover, based on the results obtained we propose a combined approach which has the potential to significantly improve the CRS stacking process.