Fifty-six individually penned pregnant ewes were offered silage ad libitum with wheat or barley in a replicated experiment. Each of these cereals was either unprocessed or processed by rolling or milling and was offered at 300 g/day from 6 weeks before lambing and at 600 g/day from 3 weeks before lambing. Five and 2 weeks before lambing, total faeces were collected over 6-day periods. On average proportionately 0·18 of the unprocessed wheat and 0·20 of the unprocessed barley consumed was recovered as intact grain in the faeces. Mean digestibility of the organic matter in the complete diet increased from 726 g/kg on whole to 748 g/kg on rolled and 759 g/kg on milled cereal (P < 0·01) at the 300 g/day level and from 721 g/kg on whole to 766 g/kg on rolled and 772 g/kg on milled cereals (P < 0·001) at the 600 g/day level. When the effect of egested whole grain was removed there was no evidence of an effect of cereal processing on organic matter digestibility and although digestibility of fibre was higher with whole than with processed grain the effect was not significant. Wheat-based diets were more digestible than barley-based diets and, compared with barley, wheat appeared to have a beneficial effect on fibre digestion.