The coenzyme NAD plays a major role in metabolism as a key redox carrier and signaling molecule but current measurement techniques cannot distinguish between different compartment pools, between free and protein-bound forms and/or between NAD(H) and NADP(H). Local free NAD/NADH ratios can be determined from product/substrate ratios of suitable near-equilibrium redox reactions but the application of this principle is often precluded by uncertainties regarding enzyme activity, localization and coenzyme specificity of dehydrogenases. In Saccharomyces cerevisiae, we circumvented these issues by expressing a bacterial mannitol-1-phosphate 5-dehydrogenase and determining the cytosolic free NAD/NADH ratio from the measured [fructose-6-phosphate]/[mannitol-1-phosphate] ratio. Under aerobic glucose-limited conditions we estimated a cytosolic free NAD/NADH ratio between 101(±14) and 320(±45), assuming the cytosolic pH is between 7.0 and 6.5, respectively. These values are more than 10-fold higher than the measured whole-cell total NAD/NADH ratio of 7.5(±2.5). Using a thermodynamic analysis of central glycolysis we demonstrate that the former are thermodynamically feasible, while the latter is not. Furthermore, we applied this novel system to study the short-term metabolic responses to perturbations. We found that the cytosolic free NAD–NADH couple became more reduced rapidly (timescale of seconds) upon a pulse of glucose (electron-donor) and that this could be reversed by the addition of acetaldehyde (electron-acceptor). In addition, these dynamics occurred without significant changes in whole-cell total NAD and NADH. This approach provides a new experimental tool for quantitative physiology and opens new possibilities in the study of energy and redox metabolism in S. cerevisiae. The same strategy should also be applicable to other microorganisms. Biotechnol. Bioeng. 2008;100: 734–743. © 2008 Wiley Periodicals, Inc.