Two in vitro experiments were carried out to quantify the incorporation of nitrogen (N) by pig colonic bacteria during the fermentation of dietary fibre, including non-starch polysaccharides and resistant starch. In the first experiment, five purified carbohydrates were used: starch (S), cellulose (C), inulin (I), pectin (P) and xylan (X). In the second experiment, three pepsin-pancreatin hydrolysed ingredients were investigated: potato, sugar-beet pulp and wheat bran. The substrates were incubated in an inoculum, prepared from fresh faeces of sows and a buffer solution providing 15 N-labelled NH 4 Cl. Gas production was monitored. Bacterial N incorporation (BNI) was estimated by measuring the incorporation of 15 N in the solid residue at halftime to asymptotic gas production (T/2). The remaining substrate was analysed for sugar content. Short-chain fatty acids (SCFA) were determined in the liquid phase. In the first experiment, the fermentation kinetics differed between the substrates. P, S and I showed higher rates of degradation (P , 0.001), while X and C showed a longer lag time and T/2. The sugar disappearance reached 0.91, 0.90, 0.81, 0.56 and 0.46, respectively, for P, I, S, C and X. Among them, S and I fixed more N per gram substrate (P , 0.05) than C, X and P (22.9 and 23.2 mg fixed N per gram fermented substrate v. 11.3, 12.3 and 9.8, respectively). Production of SCFA was the highest for the substrates with low N fixation: 562 and 565 mg/g fermented substrate for X and C v. 290 to 451 for P, I and S (P , 0.01). In the second experiment, potato and sugar-beet pulp fermented more rapidly than wheat bran (P , 0.001). Substrate disappearance at T/2 varied from 0.17 to 0.50. BNI were 18.3, 17.0 and 10.2 fixed N per gram fermented substrate, for sugar-beet pulp, potato and wheat bran, respectively, but were not statistically different. SCFA productions were the highest with wheat bran (913 mg/g fermented substrate) followed by sugar-beet pulp (641) and potato (556) (P , 0.05). The differences in N uptake by intestinal bacteria are linked to the partitioning of the substrate energy content between bacterial growth and SCFA production. This partitioning varies according to the rate of fermentation and the chemical composition of the substrate, as shown by the regression equation linking BNI to T/2 and SCFA (r 2 5 0.91, P , 0.01) and the correlation between BNI and insoluble dietary fibre (r 5 20.77, P , 0.05) when pectin was discarded from the database.