We have tated the potential improvement achieved by the C'IE94 color-diference model [ CIE Publication I16-19951 when a linear fiinction of lightness (S, = K 1 + K2 L *) is introduced as a weighting firnction for the lightness diflerence AL *. Our analyses are based on e.xperimental results previously obtained using object col0r.v and visual colorimeters; new experimental results are not reported here. For the RIT-Dupont dataset [Color Res. Appl. 16,297-316, 19911, the optimal values ofthe linear,function tested are K I = 1.02; K2 = 0.001, in excellent agreement with the proposal ofthe CIE94 model: K1 = 1.0; K2 = 0. However, jbr the dataset ofCoates et al. [ J. SOC. Dyers Colour. 97, I79-183, 19811, the best predictions were obtained using K I = 0.5; K2 = 0.008, improving the results obtained by CMC(C:c) or BFD(E:c). The.w last values difler by a,factor two from the proposal KI = 1.0; K2 = 0.017 made in another recent study [Color Res. Appl. 19, 273-276, 19941. Parametricfactors, .mch us the texture of the samples and the lightness ofthc hackgronnd, appear to be &he muin causes of the discrc.pancic~s,f~und between the best weighting fiinction for lightncw, using diflerent experimental datasets. Whilc noti experimental results are reported to clarij) this problem. the present results support the proposal made by the CIE94 model, suggesting that the lightnessdependence correction can be considered as nonrobust and related to local parametric effects. 0 1996 John Wiiev &