We propose an experimental approach combining 1H-NMR and 13C-NMR spectroscopy to investigate metabolite flux in cells under physiological conditions and present a mathematical model giving the relationships between the following different parameters. 13C fractional enrichment, fluxes in competing pathways, metabolite concentration and experimental time. This model has been used for determining the absolute and/or relative values of five fluxes involving pyruvate, ethanol, acetyl-CoA and glutamate via the Krebs cycle in glucose-grown repressed Saccharomyces cerevisiae cells fed with [1-13C]glucose and/or unlabeled ethanol. The glucose consumption and the production of various compounds such as ethanol, glycerol, trehalose etc. were studied qualitatively and/or quantitatively as a function of time. The 13C fractional enrichment of [2-13C]ethanol was determined by observing the proton resonance of the methyl group. Addition of 25 mM unlabeled ethanol shows no significant effect on the glucose consumption or the production of any metabolites. However unlabeled ethanol exerts a strong influence on the enrichment of glutamate C4, but only induces an insignificant change on glutamate C2 and C3. Apart from the fact that ethanol is a potential precursor of acetyl-CoA as expected, these results indicate that (a) the probability for citrate and 2-oxoglutarate to make one turn or more in the Krebs cycle is negligible and (b) the scrambling between C4 and C3 via the glyoxylate shunt is virtually absent. The flux of ethanol formation from pyruvate is about three-times and nine-times greater than that of ethanol consumption and acetyl-CoA formation, respectively, from pyruvate via pyruvate dehydrogenase. Without addition of unlabeled ethanol, the ratio of the integrated resonance of glutamate (C2 + C3)/C4 reflecting the activity of pyruvate carboxylase relative to that of citrate synthase, is about 1.1. By comparing the absolute values of the different fluxes, it was found that 88% of the glucose was used to synthetize ethanol but the observed concentration of ethanol in the supernatant represents only 58% of the glucose consumption. The validity of the present model was supported by the data obtained from similar experiments using unlabeled ethanol and non-NMR techniques.