The disadvantages of the traditional color image retrieval method based on color histogram are the large data storage and high complexity of computation. And what's more, the retrieval results with the condition of different illumination are not good as expected. So a fast image retrieval method based on the weighted chromaticity histogram is proposed in this paper. The method eliminates the condition of illumination using the illumination invariance model. And the image of 2-dimensional chromaticity histogram after normalization is compressed by wavelet and DCT transform, where the gray value of its pixel has been weighted based on the proportion of the chromaticity represented by the pixel in the raw image. Experiments show that retrieval results using the method presented are improved. And the time taken in retrieval process is much shorter than that when using traditional methods.