Reliable information about the seafloor and river-bed bathymetry is of high interest for a large number of applications. A Multi-Beam echo sounder (MBES) system is able to produce high-resolution bathymetry data at relatively small cost. These measurements, providing a depth for each beam and every ping, are processed to obtain a more ordered structure, such as a grid. Most approaches for assigning a depth to the centre of a cell (in a grid) use the shallowest or the mean depth in each cell. However, while the grid derived from the mean depth might be too deep compared to the shallowest depth, using the shallowest depth approach can result in an artificially shallow grid, affected by outliers. This paper introduces a number of alternatives to the current methods by combining the mean depth with statistical properties derived from the point cloud of the MBES data. In addition, the possibility of assigning a depth based on the regression coefficients of each cell is considered. The methods introduced have been tested on data acquired in different survey areas. The resulting grids have been compared to their shallowest and mean counterparts to obtain a better understanding of their advantages and limitations.