The gene encoding glucose dehydrogenase has been identified by Southern analysis of doubly restricted genomic Thermoplasma acidophilum DNA, using two redundant 17-residue oligonucleotide probes reverse translated from protein N-terminal sequence data. A 1670-bp BamH1-EcoR1 restriction fragment was ligated into pUC19 and pUC18 (constructs pTaGDH1 and pTaGDH2, respectively) and cloned in Escherichia coli. The sequence of the whole fragment was determined, and a 1059-bp open reading frame identified as the gene encoding glucose dehydrogenase. Cell-free extracts from E. coli carrying construct pTaGDH1 displayed glucose dehydrogenase activity indistinguishable from controls, but extracts from cells carrying pTaGDH2 displayed a 600-fold increase in glucose dehydrogenase activity. For high-level expression and purification of native protein, the glucose dehydrogenase coding sequence was subcloned into pMEX8. Glucose dehydrogenase purified from E. coli expressing the pMEX8 construct was indistinguishable by SDS/PAGE, N-terminal amino-acid sequence and kinetic analysis from the native enzyme purified from Tp. acidophilum. The derived 352-amino-acid sequence shows less than 20% identity with the glucose dehydrogenases of Bacillus subtilis and Bacillus megaterium but, by comparison with other eubacterial and eukaryotic dehydrogenase sequences, a portion of its sequence has been tentatively identified as a cofactor-binding region.