Fan et al. have reported that a blood urea > 7.4 mmol/L and/or glucose > 11 mmol/L at the time of admission to hospital detect severe acute pancreatitis with a sensitivity of 76% and specificity of 80.3%. We have retrospectively applied these criteria to a consecutive series of 125 patients with a diagnosis of acute pancreatitis within the West of Scotland. Utilizing the Hong Kong criteria we were only able to detect severe attacks of acute pancreatitis with a sensitivity of 33% (specificity of 83%). By lowering the concentration of urea and glucose employed to > or = 4.9 mmol/L and > or = 7.5 mmol/L, respectively, and utilizing peak concentrations recorded during the first 48 h rather than values measured at the time of admission, we were able to increase the sensitivity to 65% while reducing the specificity to 77%. Even after modification, the Hong Kong criteria were a less effective means of identifying severe attacks of acute pancreatitis than either the Glasgow scoring system (sensitivity 78% and specificity 86%) or clinical assessment at 48 h (sensitivity 82% and specificity 96%).