This paper introduces a new color texture retrieval method, based on wavelet decomposition of color texture images on the Hue/Saturation plane. The complex Hue/Saturation representation, used in the proposed method, in the Cartesian coordinates solves the problem of the traditional Hue representation in the angular coordinate and implicitly contains the correlation between the two chromatic parts of color pixel. Histograms of the extracted complex wavelet coefficients are used as features to represent the color texture images. The Jeffery divergence is used as a distance measure between the features of the query texture images and those of the images in the database. Experimental results show the new method overperforms its countparts that do not consider the correlations between color layers.