Steers, fitted with simple rumen and duodenal cannulas, were given a basal diet of approximately equal parts of straw and flaked maize [containing 12 g N kg-1 dry matter (DM)] or similar diets in which part of the flaked maize was replaced by different amounts of untreated (UT) casein or formaldehyde-treated (FT) casein to give low (L), medium (M) or high (H) intakes of nitrogen (19, 26 and 34 g N kg-1 DM, respectively). Rumen bacteria, separated from samples of rumen digesta taken about 4 h after giving a feed of the basal diet, contained approximately 160 g OLdextran-glucose ('starch') and 300 g crude protein (CP) kg-1 DM. Bacteria from steers receiving the highest level of FT or UT, contained approximately 140 and 70 g 'starch' and 320 and 580 g CP kg-l DM, respectively. It was calculated that bacteria contributed approximately 65, 50 and 30 g 'starch' day-' at the duodenum of steers receiving basal, FT casein and UT casein diets, respectively. Of the total carbohydrates entering the small intestine, all the rhamnose and ribose and approximately 50-65 % of the mannose, galactose and 'starch' were contributed by the bacteria, the remainder being dietary in origin. Virtually all the xylose, arabinose and cellulose-glucose were contributed by surviving dietary materials. For steers receiving the basal ration, mean coefficients of digestibility between mouth and duodenum (corrected for bacterial contribution) were 0.91, 0.97, 0.79, 0.43, 0.22 and 0.41 for mannose, starch-glucose, galactose, arabinose, xylose and cellulose-glucose, respectively. Coefficients of digestibility for diets supplemented with FT caseins were not appreciably different from those obtained with basal diets, but for diets supplemented with UT caseins those of arabinose, xylose and cellulose-glucose were significantly increased to 0.66, 0.55 and 0.56, respectively.