A randomised controlled trial of insulin and glucagon infusion was carried out in 18 patients in grade III or IV coma from fulminant hepatic failure due to viral or drug-induced hepatic necrosis to see whether mortality could be reduced by stimulating hepatic regeneration. Nine patients received a continuous infusion of insulin 3 U/h and glucagon 200 micrograms/h made up in 5% dextrose containing 1% human albumin solution (HAS) while controls received 5% dextrose and HAS alone. Baseline plasma insulin and glucagon levels were comparably raised in both groups and, on infusion, rose significantly higher in the insulin- and glucagon-treated patients compared to controls. Two control and one treated patient recovered. Median survival time from enrolment to death was similar for insulin- and glucagon-treated patients and controls--2 and 3 days, respectively. Insulin and glucagon therapy did not enhance hepatic synthetic function, as measured by a fall in prothrombin time or a rise in alpha-fetoprotein; nor did it stimulate hepatic regeneration, only one patient in each group showed histological evidence of hepatic regeneration at post-mortem.