The precision of measurements obtainable with several types of colorimeter (Colormaster Marks IV and V, Small Sphere Color‐Eye DI, Colorcord Mk IIA, Harrison 61 and Harrison 70) for several substrates (ceramic tiles, fabrics, yarns and threads, loose stock, slubbing and plain carpet) has been studied. Several methods of preparation and presentation of the sample to an instrument are considered and the results are compared in terms of standard deviations in 1964 (CIE) units. For textile samples the most generally satisfactory technique is to present the prepared sample behind glass in a rotating head. Under these conditions a wide range of fabrics can be measured with approximately the same precision (0<1 CIE) as ceramics. For yarns and threads, carefully wound on formers, the precision is approximately 0<3 CIE, and a slightly better precision can be obtained for carpets. Values of < 0<3 have been obtained for loose stock using a special rotating head, but values for slubbing are somewhat worse (0<8–1<0). Except for the Colorcord, good between‐instrument reproducibility (< 0<3 CIE) can be obtained for measurements of colour differences. For a given sample the absolute values (X, Y, Z) vary considerably with the instrument, the differences between pairs of instruments being equivalent to several CIE units and up to 19 when the Harrison 70 is used, even for measurements on a ceramic tile.