Primary breast cancers from 85 patients undergoing post-surgical adjuvant chemotherapy were analyzed for five glycolytic enzymes: lactate dehydrogenase (LDH); phosphohexose isomerase (PHI); glucose-6-phosphate dehydrogenase (G-6PD); pyruvate-kinase (PK); and 6-phospho-gluconate dehydrogenase (6-PGD). The purpose of this study was to determine whether biochemical parameters could offer a prognostic index to determine outcome of therapy. The patients were followed up to a maximum of 54 months; during this period 30 of them developed recurrent or metastatic disease. The enzyme activities were expressed by the three following reference parameters: units/g proteins, units/g tissue weight and units/mg DNA. Two methods of analysis were compared: firstly, univariate analysis using life tables; and secondly, multivariate analysis using the Cox's model, where enzyme levels were tested for each mode of expression in addition to node status, histological features, receptor and menopausal status. Life table analyses appear limited when subsets of patients were studied because the sample size tends to become too small to warrant firm conclusions. Using the Cox's model, a prognostic index 1 was proposed, including the number of involved nodes and the product of logarithms of G-6PD and 6-PGD expressed as units/mg DNA. Compared to the number of involved nodes, this index gives a slightly better discrimination of the patients at 2 yr after mastectomy.