ABSTRACT Clostridium thermocellum is a cellulolytic thermophile considered for consolidated bioprocessing of lignocellulose to ethanol. Improvements in ethanol yield are required for industrial implementation, but incompletely understood causes of amino acid secretion impede progress. In this study, amino acid secretion was investigated by gene deletions in ammonium-regulated NADPH-supplying and -consuming pathways and physiological characterization in cellobiose- or ammonium-limited chemostats. First, the contribution of the NADPH-supplying malate shunt was studied with strains using either the NADPH-yielding malate shunt (Δ ppdk ) or redox-independent conversion of PEP to pyruvate (Δ ppdk Δ malE::P eno -pyk ). In the latter, branched-chain amino acids, especially valine, were significantly reduced, whereas the ethanol yield increased 46-60%, suggesting that secretion of these amino acids balances NADPH surplus from the malate shunt. Unchanged amino acid secretion in Δ ppdk falsified a previous hypothesis on ammonium-regulated PEP-to-pyruvate flux redistribution. Possible involvement of another NADPH-supplier, namely NADH-dependent reduced ferredoxin:NADP + oxidoreductase ( nfnAB ), was also excluded. Finally, deletion of glutamate synthase ( gogat ) in ammonium assimilation resulted in upregulation of NADPH-linked glutamate dehydrogenase activity and decreased amino acid yields. Since gogat in C. thermocellum is putatively annotated as ferredoxin-linked, which is supported by product redistribution observed in this study, this deletion likely replaced ferredoxin with NADPH in ammonium assimilation. Overall, these findings indicate that a need to reoxidize NADPH is driving the observed amino acid secretion, likely at the expense of NADH needed for ethanol formation. This suggests that metabolic engineering strategies on simplifying redox metabolism and ammonium assimilation can contribute to increased ethanol yields. Importance Improving the ethanol yield of C. thermocellum is important for industrial implementation of this microorganism in consolidated bioprocessing. A central role of NADPH in driving amino acid byproduct formation was demonstrated, by eliminating the NADPH-supplying malate shunt and separately by changing the cofactor specificity in ammonium assimilation. With amino acid secretion diverting carbon and electrons away from ethanol, these insights are important for further metabolic engineering to reach industrial requirements on ethanol yield. This study also provides chemostat data relevant for training genome-scale metabolic models and improving the validity of their predictions, especially considering the reduced degree-of-freedom in redox metabolism of the strains generated here. In addition, this study advances fundamental understanding on mechanisms underlying amino acid secretion in cellulolytic Clostridia as well as regulation and cofactor specificity in ammonium assimilation. Together, these efforts aid development of C. thermocellum for sustainable consolidated bioprocessing of lignocellulose to ethanol with minimal pretreatment.