1. Lactating grazing ewes, fistulated at the rumen and abomasum were either not supplemented or offered, individually, 600 g/d of either pelleted, molassed sugar-beet pulp ('energy' supplement) or a pelleted 1:1 (w/w) mixture of this feed and formaldehyde-treated soya-bean meal ('protein' supplement). Digesta flows at the abomasum were estimated from the concentrations of the markers CrEDTA and ruthenium phenanthroline complex, during their administration by continuous intra-rumen infusion. Digesta samples were taken at 09.00 hours and at six further times at 4 h intervals. This was repeated 2 d later. 2. No significant difference in flow between days was noted. Daily flows of dry matter (DM) and non-ammonia-nitrogen (NAN) (g/d) in supplemented ewes were significantly higher than in unsupplemented ewes, both in total digesta and its particulate phase. However, there were significant differences between sampling times or significant interactions between sampling time and supplement treatment. The results were therefore examined by Fourier analysis for possible circadian variation in digesta flow. As there was marked between-animal variability in flow-rate, the digesta-flow values for each sampling time were re-expressed as percentages of the flows calculated from daily mean marker concentrations. 3. In unsupplemented animals, marked and significant circadian variation was then identified in the flow of DM, total N and NAN in both whole digesta and the particulate phase. There was also significant circadian variation in the flow of DM, total N and NAN in the digesta of the supplemented ewes. Curves were of the same general shape as those for unsupplemented animals, but some significant differences were found, principally for digesta NAN flow. In both supplemented and unsupplemented animals, peak flows occurred in the period 20.00-01.00 hours. The proportion of DM and NAN flowing in the particulate phase was relatively constant through the day for unsupplemented ewes, but significant circadian variation occurred for supplemented ewes. 4. The possible mechanisms generating the circadian variation in digesta flow, such as grazing behaviour and the rumen digestion of supplements, are discussed. Values are also presented to indicate the extent of the likely errors if digesta flows were estimated from samples which did not represent all stages of the circadian pattern.