Fermenting anaerobic cultures of Escherichia coli were observed by the nonintrusive technique of in vivo, whole-culture nuclear magnetic resonance. Fermentation balances were calculated for hexoses, pentoses, sugar alcohols, and sugar acids. Substrates more reduced than glucose yielded more of the highly reduced fermentation product ethanol, whereas more-oxidized substrates produced more of the less-reduced fermentation product acetate. These relationships were made more obvious by the introduction of ldhA mutations, which abolished lactate production, and Afrd mutations, which eliminated succinate. When grown anaerobically on sugar alcohols such as sorbitol, E. coli produced ethanol in excess of the amount calculated by the standard fermentation pathways. Reducing equlivalents must be recycled from formate to account for this excess of ethanol. In mutants deficient in hydrogenase (hydB), ethanol production from sorbitol was greatly decreased, implying that hydrogen gas released from formate by the formate-hydrogen lyase system may be partially recycled, in the wild type, to increase the yield of the highly reduced fermentation product ethanol.The well-known laboratory organism Escherichia coli is a facultative anaerobe. In the absence of oxygen and alternative electron acceptors such as nitrate (11), E. coli ferments sugars and their derivatives to a mixture of acids and ethanol (D. P. Clark, FEMS Microbiol. Rev., in press). This has been referred to as the mixed-acid fermentation because a mixture of acetic, lactic, formic, and succinic acids is produced (8,19). The ratio of these products has been measured by classical chemical methods for glucose fermentation under various culture conditions (7,17,20). Generally speaking, the overall proportion of lactic acid increases under acidic conditions, while the ratio of ethanol to acetate remains approximately 1:1 (1, 8). We have shown that the fermentative lactate dehydrogenase, which is responsible for lactic acid synthesis, is conjointly induced by acidic and anaerobic conditions (13). On the other hand, earlier reports that the nature of the buffer affects the fermentation product ratio (8,20) were not confirmed (13). We believe that the difference in buffering strength and pK between various buffers may give the illusory appearance of such an effect. However, when these factors are allowed for, only the overall pH maintained is significant (13).A factor which greatly affects the fermentation product ratio is the redox level of the substrate relative to glucose. Thus, sugar alcohols, which are more reduced than the corresponding hexoses, must yield a higher proportion of more-reduced fermentation products in order to achieve hydrogen balance (Clark, in press). As far as we know, no systematic survey has been done to relate the redox level of the fermentable substrate to the composition of the fermentation product. Here we report such a study based on the use of in vivo nuclear magnetic resonance (NMR) of anaerobic cultures (15, 16). The use of in vivo NMR avoids...