Summary. Automatization of a sensitive fructose assay used in the study of soya oligosides.We report the automatization of a sensitive specific assay of free fructose or that contained in various molecules. The manifold is shown in figure 1 The method is used to specifically assay saccharose and the a-galactosides (raffinose, stachyose) of various soya samples (oil meal, fermented oil meal, flour, fermented flour, soya feed) after separation by resin anion-exchange chromatography. These sugars are assayed at the same time using a total sugar evaluation method (sulfuric orcinol). Two methods for determinating fructose concentration in the column effluent are compared. One uses direct assay at the column exit at the same time as the total sugar is assayed. Oligoside concentration is obtained by comparison with the standard curves S = f(C) (S = surface, C = concentration) for each sugar. In the other method, the part of the effluent not used in total sugar assay is collected and the fructose is assayed on each carbohydratfraction. Multiplying the result by a constant measured for each sugar, oligoside concentration in the sample may be determined.Results obtained by both fructose assay methods agree and they confirm those found by assaying the total sugar (table 3). Differences between the methods are non-significant (P G 0.05). Thus, the peaks separated by the resin column contain no other carbohydrated which could interfere with the saccharose and the a-galactosides. If they did, the unseparated carbohydrates would be evaluated by difference. The results on the carbohydrate composition of soya agree with those found by other authors. Fermentation eliminates a large part of the carbohydrates from the samples. The choice of a method for evaluating saccharose and a-galactoside, using fructose, is discussed. The second is recommended because the curves S = f(C) vary in time ; this may cause errors.Introduction.