Two experiments were carried out on cattle nourished entirely by intragastric infusion, to determine the extent to which glucose or a glucose precursor determines the response to protein infusion in energy-undernourished animals. In order to determine the requirement for glucose in 1-year-old fasting cattle, glucose was infused at increments to supply 0, 1·5, 2·5, 3·5, 4·5, 5·5 and 6·5 g/kg metabolic body weight (W0·75) and the effects on plasma β-hydroxybutyrate and N excretion were measured. At 5·5 g glucose/kg W0·75 plasma β-hydroxybutyrate was reduced to a basal level of 1·65 mmol/l and fasting N excretion reduced from 529 to 280 mg N/kg W0·75. No further reduction was observed with the higher level of 6·5 g glucose/kg W0·75. In the second trial, three steers were used in a 3 × 3 Latin square design and infused with a volatile fatty acid mixture of 65, 27 and 8 mol acetic, propionic and butyric acids respectively/100 mol, either at an estimated maintenance energy level of 450 kJ/kg W0·75 and supplying a calculated glucose equivalent level of 13·0 g/kg W0·75 (M1A), or at 1·5 × maintenance supplying a glucose equivalent of 20 g/kg W0·75 (M1·5A). Another mixture infused at the maintenance energy level contained 49, 43 and 8 mol acetic, propionic and butyric acids respectively/100 mol but with a glucose equivalent of 20 g/kg W0·75 (M1P). Casein was infused at each of these energy treatments to supply 0, 200, 400, 800, 1600 and 2500 mg N/kg W0·75 daily, and N balance and blood metabolites were measured. N retention increased linearly (r 0·98) with casein infusion. The coefficients for N retention were 0·55, 0·57 and 0·64 for M1A, M1·5A and M1P respectively. The mean efficiency of N utilization was 0·58. The results suggest that provided the glucose need is met there is no relationship between energy supply and efficiency and level of protein retention. However, the results also indicate that glucose requirement in cattle may be higher than that previously observed in sheep.