In a previous report we showed that cholestasis in diabetic rats is due in part to hyperglycemia. To gain information about the mechanism responsible for this phenomenon, bile flow was studied in isolated perfused rat livers. The perfusion media were modified erythrocyte-free Krebs-Henseleit solutions. Under these experimental conditions, no cholestasis was observed in isolated rat livers obtained from rats treated with streptozotocin (6 mg/100 gm body wt) 6 days before the experiments. We then proceeded to use normal animals. The composition of the perfusion media was modified to maintain the osmolality even after increasing D-glucose concentrations from 0 to 35 mmol/L. Bile flow was not affected with doses up to 15 mmol/L D-glucose. Beyond a threshold value for plasma D-glucose concentrations between 15 and 20 mmol/L, cholestasis was observed. Using D-glucose analogs such as L-glucose and 3-O-methyl-D-glucose, bile flow was also reduced (by 0.54 and 0.53 microliters/min/gm liver, respectively). Isosmotic sucrose-containing perfusion media were also observed to impair bile flow (by 0.66 microliters/min/gm liver). However, i-erythritol and D-mannitol failed to inhibit bile formation. The study of bile/plasma concentration ratios determined using tracer amounts of radioactive sugars indicated that this value was much lower for cholestatic sucrose (0.11) and L-glucose (0.31) than for noncholestatic i-erythritol (0.99) and D-mannitol (0.98). Cholestasis was partly reversed after induction by 35 mmol/L D-glucose if perfusion media were replaced by sugar-free ones, but also by media containing 25 mmol/L D-glucose. Insulin given during the perfusion with sugar-free media was observed to have no effect on bile flow.(ABSTRACT TRUNCATED AT 250 WORDS)