Modifications have been proposed in an attempt to improve the clinical value of the original nine-factor Glasgow prognostic scoring system for acute pancreatitis. These include the omission of age or serum transaminase, reducing the factors to eight. Debate exists as to which system should be employed. Assessment of the individual factors in 198 attacks of acute pancreatitis treated conventionally revealed that only serum transaminase did not differ significantly between mild and severe outcome groups. Multivariate analysis demonstrated four factors (Pa,O2, white cell count, lactic dehydrogenase, and urea) with independent significance in predicting severity, while serum glucose, albumin and transaminase were least useful. Findings were similar when considering only patients with gallstone aetiology. The reduction of the prognostic factors to eight by the omission of either age or transaminase improved the predictive value of the scoring system, both when considering all attacks and those of gallstone aetiology alone. We suggest that serum transaminase should be omitted because: it shows no significant difference between mild and severe outcome groups, while age has prognostic significance; the system has greater sensitivity than if age is omitted; and the number of factors requiring emergency laboratory measurement for immediate prognostication is reduced by one.