Ocean modellers use bathymetric datasets like ETOPO5 and ETOPO2 to represent the ocean bottom topography. The former dataset is based on digitization of depth contours greater than 200 m, and the latter is based on satellite altimetry. Hence, they are not always reliable in shallow regions. An improved shelf bathymetry for the Indian Ocean region (20°E to 112°E and 38°S to 32°N) is derived by digitizing the depth contours and sounding depths less than 200 m from the hydrographic charts published by the National Hydrographic Office, India. The digitized data are then gridded and used to modify the existing ETOPO5 and ETOPO2 datasets for depths less than 200 m. In combining the digitized data with the original ETOPO dataset, we apply an appropriate blending technique near the 200 m contour to ensure smooth merging of the datasets. Using the modified ETOPO5, we demonstrate that the original ETOPO5 is indeed inaccurate in depths of less than 200 m and has features that are not actually present on the ocean bottom. Though the present version of ETOPO2 (ETOPO2v2) is a better bathymetry compared to its earlier versions, there are still differences between the ETOPO2v2 and the modified ETOPO2. We assess the improvements of these bathymetric grids with the performance of existing models of tidal circulation and tsunami propagation.