A method for the quantitative analysis of casein mixtures is described. The method uses ion-exchange chromatography at 4 °C on DEAE cellulose, Whatman DE52 in the presence of tris-chloride-urea buffer (pH 8-6), and a NaCl gradient to fractionate the alkylated casein mixtures and a micro-biuret technique to determine protein concentrations. Values for fractions representative of (1) y-caseins, (2) /c-casein plus some unknown proteins, (3) /?-casein, (4) minor a g -caseins and (5) a 8l -plus a 80 -caseins were obtained, and whole casein from herd bulk milk was found to consist on average of about 2, 13, 36, 11 and 38% respectively of these fractions. Agreement between duplicate fractionations was satisfactory and the recovery of material varied between 94 and 98%.Rose, Davies & Yaguchi (1969) described a method for the quantitative determination of the major components of casein mixtures which was based on the use of column chromatography on DEAE cellulose. The method gave a more extensive fractionation than that obtained by moving-boundary electrophoresis (Rolleri, Larson & Touchberry, 1956;Larson & Kendall, 1957), and yielded results which were more accurate than those derived from sialic acid (Marier, Tessier & Rose, 1963) and turbidimetric measurements (Tessier, Rose & Marier 1963) and also probably more reliable than those obtained from scanning patterns obtained by gel electrophoresis (Morr, Lin & Josephson, 1971). In the method of Rose et al. fractionation was done on DEAE cellulose, Whatman DE11, an anion exchanger with low proteinbinding capacity and relatively poor kinetic properties which has now been superseded by more advanced ion-exchange materials. In an attempt to improve on their method the improved exchanger Whatman DE52 has been used for casein fractionation and certain aspects of the procedure, such as the methods used for protein determination and for the measurement of the recovery of material, have been more critically examined. These studies have produced a system giving greater resolution of casein mixtures than achieved previously and have shown that agreement between duplicate fractionations and recovery of material is satisfactory. However, complete separation of /c-casein from other proteins was not possible and values for /c-casein are therefore somewhat overestimated.