Three diets were formulated according to the nature of the fibrous components: a concentrate low fibre diet enriched in starch (diet S, neutral-detergent fibre (NDF) proportionately 0·078 of dry matter (DM)) and two high-fibre diets enriched in either sugar-beet pulp (diet BP, NDF = 0·219 of DM) or wheat bran (diet WB, NDF = 0·187 of DM). In a first experiment, the digestibility of nutrients in these diets were determined using dry sows. In a second experiment, 24 multiparous Large White sows were offered the experimental diets in a 3 x 3 Latin-square design during three 21-day periods over pregnancy in order to evaluate physiological and behavioural effects over the nycthemeral period. On the basis of the first experiment, the daily food supply was adjusted to 2·44, 2·74 and 2-90 kg/day for diets S, BP and WB, respectively, to provide the same amount of metabolizable energy to each treatment. In the hour prior to the meal, the diet effect was nil on plasma levels of glucose, insulin, glucagon and non-esterified fatty acid and limited on cortisol plasma level. After the food delivery, the peak responses of glucose and insulin to the meal were delayed in sows given the BP diet, while their level showed a higher increase for S than for WB and BP diets. Sows offered diet BP spent less time standing than sows offered diet S. The occurrence of non-feeding oral activities over the 45 min following food distribution was significantly reduced with fibrous diets. But no difference appeared during the 45-min period following the end of each meal. It was concluded that the incorporation of beet pulp in the diet was more effective than wheat bran in extending the feeding activity and the absorption of nutrients. The diet effect was limited to the 2 h after the food delivery for these sows given a restricted amount of food.