The establishment of a reliable large animal model of acute hepatic failure (AHF) is critical for the evaluation of supportive therapies such as bioartificial liver support systems (BALSS). Large animal surgical models bear some resemblance to the clinical syndrome of AHF in humans. However, these are inherently complex and are predisposed to inter-individual variation. The development of considerable skill is required and no precise standardization criteria for such models have been defined as yet. This study investigates a surgically induced ischemic model of AHF in 15 female Landrace pigs. A large set of systemic and biochemical variables were measured. The absolute values of systemic variables during the surgery and the rates of change of the biochemical variables following the surgery were correlated with the duration of survival of each animal. A multivariate prognosis was revealed, with several variables simultaneously determining survival. Those of importance in the surgical period included the duration of portal occlusion, the pulse rate during this time, the total blood lost, the mean arterial pressure and the body temperature after the procedure. In the post-surgical period the rates of change of blood ammonia, branch chain amino acids, haemoglobin, hematocrit, body temperature, total urinary excretion and mean blood pH all demonstrated some importance to survival. Based on the above and clinical experience, standardization criteria specific for this model were defined. Additional variables that posses value in AHF but demonstrate lower correlations with survival were also included, e.g. the prothrombin time, clotting factors, liver enzymes, potassium, bilirubin, creatinine and lactate. Valuable procedural information was generated, which stabilized the model, limited unnecessary blood sampling and cut laboratory costs. The criteria allow the early exclusion of compromized animals and the identification of those demonstrating characteristics predictive of longer survival prior to BALSS connection. This is likely to allow more accurate comparisons between small treatment and control groups. In effect, the above are the converse of the King's college criteria (for humans) as they are designed to exclude animals with prognoses that are insufficiently bleak or due to factors other than AHF. The statistical analysis employed also identified the variables that in future may allow the evaluation of prognosis in real-time.