An experiment with ruminally cannulated sheep was undertaken to determine how the dietary energy source affected the ability of rumen fermentation to respond to pre-formed amino acids compared to non-protein nitrogen in the diet. Four sheep received diets of grass hay (H) or molassed sugar beet pulp (BP) supplemented with 9.3 g N kg DM added either as urea (U) or casein (C) in a 4 × 4 latin square with a factorial design. Measurements of rumen fermentation, microbial numbers and purine excretion were made in the third week of each 21-day period. Rumen total volatile fatty acids concentrations were higher with the BP diets (130 vs. 91 mM; P <0.01), and the molar proportion of propionate was lower and that of butyrate was higher with BP, but the nitrogen source had no significant effect on volatile fatty acids. Rumen ammonia concentrations were lower with the BPU diet than with HU, and were decreased when casein replaced urea in both diets (11.1, 19.4, 5.0 and 10.2 mM for BPU, HU, BPC and HC respectively; P <0.01). The source of nitrogen had no influence on microbial numbers or yield in the hay diets, but total viable bacteria more than doubled in response to pre-formed amino acids with BP (16.9 vs. 7.4 × 10 ml; P <0.01). Microbial protein flow, calculated from urinary purine excretion, also increased in response to casein but only with BP (13.5 vs. 7.4 g N per day). The rate of loss of diets from nylon bags was much more rapid (P <0.01) with BP than with hay, and was unaffected by dietary nitrogen source (P > 0.05). A microbial response to rumen-degradable protein therefore occurred only with the more rapidly degraded BP diet, consistent with the idea that rumen microbial growth responds to pre-formed amino acids only when the energy source is fermented rapidly.