An anisotropic prestack depth-migration sequence was used in the imaging of a carbonate accumulation in deep water offshore Brazil. The reservoir is overlain by a complex sequence of evaporites, carbonates, and clastics. Although the latter display a modest magnitude of anisotropy, this parameter has to be taken into account for the depth accuracy required in reservoir evaluation. The seismic data input to the processing consist of two narrow-azimuth streamer surveys, acquired with distinct azimuths that overlap approximately at the center of the reservoir structure. The resulting velocity model is geologically plausible, exhibiting good accuracy when compared with average velocities calculated from check shots and vertical seismic profiles acquired at nearby wells. Such accuracy holds true to the base of the evaporitic sequence, which corresponds roughly to the top of reservoir, and makes this model adequate for depth positioning and hydrocarbon volume calculations. However, deep layers in the velocity models present larger errors and a spatial behavior that does not match geologic expectations. This is attributed to the lack of proper illumination that these conventional seismic acquisitions provide and arguably makes the case for a more modern full-azimuth seismic acquisition over the structure. This limitation is quantified by an illumination map simulated at the top of the reservoir from a waveform-modeling exercise that took into consideration the seismic-acquisition parameters over the field.