Ten color‐difference formulae were evaluated by means of a psychophysical experiment on perception scales of color attributes. Mattiello and Guirao1,2have worked on scaling of surface colors. The present work uses the data obtained by them to check the behavior of different color‐difference formulae.The color differences between samples used in these experiments are large in comparison with the just‐perceptible or threshold criterion which has been considered as the most important from the industrial point of view. Nevertheless, the preliminary information collected shows that results obtained by other authors for small color differences are confirmed for the present type of color differences.The present work has two parts: the first deals with lightness variations of color samples, with hue and saturation approximately constant. The second part analyzes result for samples in which saturation was varied, maintaining the other two variables almost constant.For the first part, the Reilly Cube Root formula has the highest correlation coefficient, followed by the CIEL*a*b*,ANLAB, and FMC‐2 formulae, in decreasing order.For the second part, the Judd—Hunter (NBS) formula has the highest correlation coefficient, followed by the Hunter—Scofield, Reilly Cube Root, and CIEL*a*b*formulae.For these results, and taking into account the limitations of color‐difference size mentioned, the recommendations proposed by the CIE TC‐1.3 Committee seem to be the best presently available.It is also possible to deduce, in view of the evidence shown for small color differences in different field trials of various formula and the present work, that the observers' behavior when seeing small or large differences in color of opaque materials seems to be simila