Two sensory difference tests have been used to assess the ability of an untrained population to perceive colour difference in a cosmetic product. The two tests used were the triangle test and the "two-out-of-five" test. Participants were presented with groups of samples with varying colour differences and asked to identify the odd sample in the triangle test and the pair in the two-out-of-five test. From these data, the number of correct responses was correlated with the calculated colour difference using three colour-difference equations (CMC, CIE94, and CIEDE2000). These correlations were optimized by varying the parameters in the colour-difference equations. With the parameters optimized, each of the three colour-difference equations gave a correlation coefficient of similar to0.97 with the two-out-of-five test and a correlation coefficient of similar to0.79 with the triangle test. These correlation coefficients suggest that sensory difference testing can be used to investigate perception of colour difference. However, for the triangle test the correlation between the sensory data and the calculated colour difference is weak and the two-out-of-five test should be preferred. The minimum perceptible colour difference was estimated from the regression plots between the optimised colour-difference equations and the sensory data. (C) 2004 Wiley Periodicals, Inc.