The present study compares estimates of rumen microbial-N production derived from duodenal flow measurements (15N and purine bases) with those from measurements of the urinary excretion of purine derivatives. Four Rasa Aragonesa ewes fitted with simple cannulas in the rumen and proximal duodenum were used. Four diets consisting of 550 g lucerne (Medicago sativa) hay/d as sole feed or supplemented with 220, 400 and 550 g rolled barley grain/d were given in a 4 x 4 random factorial arrangement. Duodenal digesta flows were determined by the dual-phase marker technique during continuous intraruminal infusions of Co-EDTA and Yb-acetate. Microbial contribution to the non-NH3N (NAN)flow was estimated from 15N enrichment and purines: N ratio in duodenal digesta and bacterial fractions isolated from the rumen content. Whole tract organic matter (OM) digestibility and duodenal flow of OM and NAN increased (P<0·001) with the level of barley supplementation. Digestible OM intake ranged from 19·0 to 42·7 g/kg metabolic weight (W0·75) and the duodenal flow of purine bases and the urinary excretion of allantoin increased Linearly (P < 0·001) from minimum values of 7·47 (SD 1·524)and 4·65 (SD 0·705) mmol/d respectively on the basal diet to 18·20 (SD 1·751) and 11·62 (SD 0·214) mmol/d on the 400 g barley diet; a further increase in barley supplementation decreased both variables (13/50 (SD 2/334) and 8/77 (SD 0/617) mmol/d respectively). Urinary excretion of uric acid and hypoxanthine showed a slight but significant increase (P < 0·05) over all levels of barley. Molar recoveries of duodenal purine bases as purine derivatives or allantoin in the urine were 0·78 (SD 0·156) and 0·65 (SD 0·130) respectively. The increase on barley supplementation significantly augmented microbial-N, but large differences between microbial markers employed were observed. Mean values of microbial-N estimated from the duodenal purine bases or urinary allantoin excretion were on average 18 and 29% lower than those measured by 15N.