A new indexing technique for image databases integrating color and similarity information for chromaticity based image retrieval is proposed. This methodology, called chromatic histogram, extracts few values of information about the location of pixels having the same color and their arrangement position within the image, can be more satisfactory than other standard techniques like text, feature, and shape when the user would like to retrieve from the image databases that actually resemble the query image selected in their color distribution properties. Experimental trials on a database of about 1000 images are implemented on the basis of human perceptual judgments.