Multivaried and trend analyses were applied to a set of 27 periodic haematochemical checks on 37 patients given surgery for biliopancreatic pathology. Preoperative and postoperative data were separately analysed. The software employed was specially designed for the purpose. The discriminating function was used in the preoperative period to establish the best moment for diagnosis, which turns out to be 5 days after admission to hospital. Variance and regression analyses were used for short term prognosis by calculating survival chances. In one case the prognosis turned out to be distinctly inaccurate.