Pentobarbitone may not be an appropriate anaesthetic to use on the donor rat of an isolated liver perfusion model intended to study drug-drug interactions, since the anaesthetic was found to remain in the liver in concentrations which may be high enough to modify the disposition of a test drug. A retrospective study to assess the anaesthetic pentobarbitone decay in six perfused rat livers, showed a linear correlation between perfusion medium concentrations of phenytoin and pentobarbitone at the early sample times, suggesting that the anaesthetic was inhibiting the hepatic clearance of phenytoin. Direct evidence for this inhibition was obtained following the addition of pentobarbitone (5 mg) to a standard perfusion experiment with phenytoin at t = 30 min. At the end of these experiments (t = 120 min), the perfusate phenytoin levels had decreased to only 18.6 +/- 3.7% of the initial concentration, compared to 5.6 +/- 2.5% in the control group. Pithing, under light ether anaesthesia, was investigated as an alternative preparatory procedure for the donor rat with potentially less effect on the liver. Phenytoin was found to be cleared from the perfusion medium more rapidly with livers thus prepared, especially in the early stages. Also the precision of the technique was improved; the standard errors for six replicate experiments being considerably less in the pithed group compared to the pentobarbitone anaesthetised group. This is clearly a more appropriate preparatory procedure for rat liver perfusion techniques designed to investigate drug metabolism. There may also be some advantages when the technique is to be used for other investigations.