Several applications fields, like remote sensing, medical science, multimedia, requires the interpretation, archivation and transmission of image data. The volume of data to be managed is very large and this stimulated the research activities aiming at image compression. A particular topic of high interest is the evaluation of the specific degradations introduced by compression. In general such studies use only the mean square error and the visual inspection to determine the quality of the reconstructed image. This criteria are not sufficient for an accurate evaluation of the reconstructed image accuracy. In this paper it is introduced a new method for the quality evaluation of the block transform compressed images. It reflects the objective function of the compression algorithm, gives a measure of the visual quality and it is easily computed.