The use of prestack interpolation prior to prestack migration to improve AVO analysis on image gathers is demonstrated on an exploration play. The interpolation achieves this improvement by reducing migration artifacts. AVO analysis attempts to estimate fundamental information from surface seismic data and likely will be used more frequently if the estimates can be more accurately produced. Land 3D seismic typically has poor and irregular sampling. This poor sampling creates migration noise, which is a material cause of inaccurate AVO estimates. Prestack 5D interpolation is applied prior to prestack migration and AVO analysis on the imaged gathers to address this noise problem. The interpolation algorithm includes offset and azimuth dimensions that preserve AVO information. This method is evaluated bycomparing the results to those of alternate approaches, such as superbinning, that suppress this kind of noise in AVO analysis. The evaluation is determined by comparing our ability to predict the reservoir quality of a gas-charged sandstone reservoir with 48 well penetrations. We compare migrated gathers, AVO attribute stacks, and attribute maps in our analysis. We also generate scatter plots of the AVO attribute values against measures of reservoir quality at the well control points to allow a quantitative measure of the improvements. The interpolation method yields gathers, stacks, and maps that all appear to be better resolved and less noisy than the other methods. The scatter plots demonstrate a measurable and significant improvement from the interpolation method, especially compared to superbinning. This work suggests interpolation before imaging, and imaging before AVO analysis, should be performed on land 3D surface seismic data.