Early image retrieval techniques were based on textual annotation of images. Annotating images manually is a cumbersome and expensive task for large image databases, and is often subjective, context-sensitive and incomplete. Content based image retrieval, uses the visual contents of an image such as color, shape, texture, and spatial layout to represent and index the image. The Region Based Image Retrieval (RBIR) system uses the Discrete Wavelet Transform (DWT) and a Watershed Segmentation algorithm to segment an image into regions. Though k-means clustering algorithm is very fast and simple to implement, it provides only coarse image segmentation. Better retrieval results can be expected by employing a more sophisticated segmentation technique. For this purpose, a novel Texture Gradient based Watershed Segmentation technique is developed. The Watershed Transform is a well established tool for the segmentation of images. However, it is often not effective for textured image regions that are perceptually homogeneous. In order to properly segment such regions the concept of the Texture Gradient is introduced and is implemented using a Non Decimated Wavelet Packet Transform. A marker location algorithm is subsequently used to locate significant homogeneous textured or non textured regions. A marker driven Watershed Transform is then used to properly segment the identified regions. The experimental results demonstrate the superiority of this technique over k-means clustering.