Four steers were maintained wholly by intragastric infusion of volatile fatty acids (VFA) and protein, together with a mineral–vitamin supplement. The infusion was given at three levels of energy, namely 450, 675 and 900 kJ/g live weight0·75, calculated to supply energy at 1·0,1·5 or 2·0 times that required for maintenance. The VFA provided 0·837 and the protein 0·163 of the energy infused. The molar proportions of individual VFA were varied so that the infusate contained 0·36–0·91 of acetic acid, 0·56–0·01 of propionic acid and a constant 0·08 of butyric acid. Heat production was measured in respiration chambers. Urine was analysed for N, urea, β-hydroxybutyrate and VFA. Blood plasma was analysed for β-hydroxybutyrate, free fatty acids, insulin and glucose. As the proportion of acetic acid was increased, and propionic acid reduced, there was no change in blood or urine metabolites or in heat production until acetic acid exceeded a proportion of about 0·75. At higher proportions β-hydroxybutyrate increased in plasma and urine, blood glucose and insulin tended to fall and urinary N excretion rose. At a proportion of acetic acid of > 0·80, acetate appeared in the urine and at > 0·86 heat production declined. The effect of level of infusion on the molar proportion at which plasma and urine metabolites changed was less clear. There was a tendency for the increase in β-hydroxybutyrate to occur at a slightly lower proportion of acetic acid at the highest level of infusion. It is concluded that differences in heat production that are observed between diets are probably not caused by differences in rumen VFA proportions. The reaction to a highly elevated proportion of acetic acid is to excrete β-hydroxybutyrate and acetate in the urine and so decrease rather than increase heat production. Regardless of level of infusion a metabolic crisis occurred when the proportion of acetic acid was above the levels found in the rumen content of normally-fed animals.