This paper is the continuation of a previous paper (Malcotti and Biondi, 1997) in which we discussed the theoretical and practical bases of depth focusing analysis for 2-D and 3-D. In this work, we present some results of the depth focusing analysis methodology based on a 3-D prestack common-azimuth migration (Biondi et. al., 1996). We apply this methodology to two differents data sets with complex velocities, a synthetic data set and a real 3-D data set. In both data sets, we show one iteration in the methodology loop for interval velocity estimation using depth focusing analysis. For the real seismic data set, our goal is to update the velocity model where the image has focusing problems. Therefore, we used the best velocity field available. For the interpretation of the depth error gathers, we use a three dimensional graphic tool that help to simplify the picking in complex error gathers.