Four ewes fitted with ruminal and duodenal T-piece cannulas were given four diets in a 4 X 4 factorial design. Diets consisted of 700 (HF) or 400 (LF) g/day of ammonia-treated barley straw supplemented respectively with 150 or 600 g/day of concentrate made up with barley plus either soya-bean meal (SBM) or fish meal (FM) as the protein source, offered at 2-h intervals. Duodenal flows of digesta were estimated by the dual-phase technique using Co-EDTA and Yb-acetate as markers and ( 15 NH 4 ) 2 SO 4 was infused into the rumen to label microbial N. Bacteria were isolated from the liquid (LAB) or solid (SAB) rumen digesta. Purine bases (PB) were isolated by precipitation in an acid solution ofAgN0 3 , and microbial contribution either to the duodenal nitrogen (N) or PB were determined by 15 N measurements in duodenal digesta and bacteria. Simultaneously, the rumen degradation of N and PB contained in SBM and FM was studied by incubating supplements in polyester bags in the rumen. PB content (jimol/g dry matter) and guanine: adenine (G/A) ratio of barley straw was barley grain, SBM, and FM, respectively. Duodenal flow ofPB (mmol/day) was significantly higher than PB intake on all diets and G/A ratio showed a mean value of 0-97, similar to the ratios determined in SAB (0-80) and LAB (1-04) and much lower than diets (1-31 to 4-32). Microbial contribution to duodenal N flow ranged from 43-3% to 61-0%, being higher in SBM (59-0%) than in FM (46-7%) diets. However, microbial contribution to duodenal PB was not affected by the experimental treatment, accounting for proportionately 0-77 of total PB at the duodenum. Rumen degradability of PB was much higher than that of total N and in both cases degradability was higher in SBM than FM. Direct measurements of non-microbial N were significantly higher than values determined by the polyester-bag measurements. However, once corrected for the endogenous N (52 mgN per kg live weight) contribution, results showed an acceptable agreement. Duodenal flow of PB non-attributable to microbes (unlabelled PB) showed a mean value of 3-25 mmol/day without a significant effect of dietary treatment. However, undegradable PB supply determined for 0-02, 0-05 and 0-08 per h fractional outflow rates were proportionately lower than 0-025 with SBM and 0-100 with FM diets of the estimated duodenal PB flow. Despite the magnitude of the unlabelled duodenal PB, the close agreement between G/A ratios in duodenal digesta and bacteria suggests that the contribution of dietary PB to the duodenal flow was low and seems to confirm the reliability of values obtained from polyester-bag measurements.