The pharmacokinetics of etomidate were studied in 9 control subjects (with normal liver function) and in 5 patients with cirrhosis scheduled for gastro-intestinal surgery. Anaesthetic induction included an initial bolus of etomidate 0.3 mg.kg-1, together with fentanyl 2 micrograms.kg-1, and pancuronium 60 micrograms.kg-1. An etomidate infusion was then started according to one of two following schemes: a (0.03 mg.kg-1.min-1 for 10 min, and then 0.01 mg.kg-1.min-1), or B (0.1 mg.kg-1.min-1 for 10 min, followed by 0.02 mg.kg-1.min-1 for a further 110 min, and 0.01 mg.kg-1.min-1 thereafter). Plasma concentrations of etomidate were determined at regular intervals throughout anaesthesia, and up to four hours afterwards, using inverse phase high pressure liquid chromatography. The infusion was given for 273 +/- 87 min in controls, and for 259 +/- 56 min in the cirrhotic group. Scheme A, only used in 3 controls and 1 cirrhotic in a preliminary study, resulted in very low plasma concentrations: 0.2 to 0.4 micrograms.ml-1. Those measured during the apparent plateau phase (steady state) of infusion protocol B were close to predicted values (0.5 to 0.6 micrograms.ml-1) in controls, whereas higher concentrations (approximately 1.5 micrograms.ml-1) were reached in cirrhotic patients. For all the patients the time interval to spontaneous recovery was 41 +/- 27 min; plasma levels were then 0.199 +/- 0.092 micrograms.ml-1. There were significant alterations in pharmacokinetic parameters in the cirrhotic patients.(ABSTRACT TRUNCATED AT 250 WORDS)