A total of 15 rumen-cannulated sheep and four rumen-cannulated goats (Expt 3) were used in five latin square experiments designed to investigate the effects of carbohydrate supplements on ruminal ammonia concentration in animals given grass silage diets. In Expt 1 barley supplements were given at the same time, 1 h before or 2 h before a meal of silage. The treatments were designed to alter the synchronization between energy release from fermentation of barley and ammonia release from the degradation of silage N compounds. As compared with the unsupplemented control diet, barley supplements reduced (P < 0–05) rumen ammonia concentration and increased (P < 0–01) the number of total protozoa, but the time at which barley was given was without effect. In Expt 2 animals receiving silage or silage-barley diets were defaunated chemically. This treatment led to a 20–25 % reduction in rumen ammonia concentration. In Expt 3 supplements of maize starch, glucose and sucrose were compared. Mean ammonia concentrations were 231 mg/1 for the control unsupplemented diet and 205, 155 and 160 mg/1 (S.E. 21) for the starch, glucose and sucrose treatments. Corresponding numbers of protozoa were 7–7, 15–1, 6–1 and 6–3 x 10s/ml (S.E. 1–8). In Expt 4 the diets were unsupplemented silage or the same diet plus supplements of sucrose or xylose. Xylose reduced ammonia concentration more than sucrose, the values (mg/1) being 236 (control), 206 (sucrose) and 125 (xylose) (S.E. 19). There was no difference between the supplements in numbers of protozoa. Xylose induced a smaller reduction in pH, a higher (P < 0–05) proportion of acetate and a lower (P < 0–05) proportion of butyrate in rumen fluid than did sucrose. Effects of rumen pH were examined in Expt 5 where supplements of sucrose were given alone or together with NaHCO3. Rumen pH values were 6–38 for the control unsupplemented diet and 5–99, 6–28 and 6–55 (S.E. 0–06) for the diets supplemented with sucrose, sucrose plus 50 g NaHC03 and sucrose plus 100 g NaHCO3. Corresponding values for ruminal ammonia were 193, 151, 93 and 41 mg/1 (S.E. 10). Differences in VFA proportions between sucrose treatments were small and significant (P < 0–05) only for butyrate. It is concluded that there are important differences between carbohydrate sources in their effects on nitrogen metabolism in the rumen. Differences between starch and sugars appear to relate to the influence of the carbohydrates on the microbial population of the rumen, as was indicated by the differential effects of the carbohydrate sources on the number of total protozoa; differences between sugars appear to depend in part on the rates of sugar fermentation and the associated reduction in rumen pH.