The aim of this work was to study the relationship between the reactions obtained by indirect methods of testing milk samples and the cell content, determined by direct cell counting after Prescott & Breed (1910). The indirect tests used were the California Mastitis Test (CMT), Whiteside Test (WsT), Brabant Mastitis Reaction (BMR), and the Catalase Test (CT). The results of the investigations are summarized below. Samples with cell values lying within fixed limits do not neces¬sarily conform to particular indirect test reactions. All four indirect tests have wide ranges of variation of cell level within a given reaction with considerable overlapping between reac¬tions. The standard deviation for cell levels within a given reaction is large, although this can be partly explained by the fact that the cell values are estimated rather than real. Correction for this produces only an insignificant decrease, however. The coefficient of correlation is a little higher for the correlation CMT—CC than for the other indirect tests and CC. When sample groups with different mean cell levels (calculated from the whole material) are compared, the variation ranges (dif¬ference between the largest and the smallest observation) and standard deviations arc generally the same for the different CMT reactions. However, the mean cell values, for different reactions lie at higher levels when the mean cell content of the whole material is higher. Similarly they are lower when the mean of the whole material is lower. This is, however, not true for 0 reactions. The mean cell values for CMT 4- and for 4- 4- reactions varied considerably in different herds. It was not considered acceptable, for any of the four indirect tests, to substitute reactions by average cell values, either to determine the mean cell value or to decide whether a sample is inacceptable, if this means that the sample has a cell content above a certain value. Negative and 4- reactions for CMT and WsT could possibly be replaced by a mean cell level, since the range of variation is not so wide here and the majority of samples giving these reactions contain less than 3xl05 cells per ml. The possibility exists of combining indirect tests and direct cell counting to examine milk samples, so that samples with low reactions could be passed, while those with higher reactions would be examined by cell counting. The indirect tests differentiate milk at different cell levels. The possibility of using a particular indirect test depends on the cell level to be examined and on the degree of accuracy required.