Bacteria which can grow in different environments have developed regulatory systems which allow them to exploit specific habitats to their best advantage. In the facultative anaerobe Escherichia coli two transcriptional regulators controlling independent networks of oxygen-regulated gene expression have been identified. One is a two-component sensor-regulator system (ArcB-A), which represses a wide variety of aerobic enzymes under anaerobic conditions. The other is FNR, the transcriptional regulator which is essential for expressing anaerobic respiratory processes. The purpose of this review is to summarize what is known about FNR. The fnr gene was initially defined by the isolation of some pleiotropic mutants which characteristically lacked the ability to use fumarate and nitrate as reducible substrates for supporting anaerobic growth and several other anaerobic respiratory functions. Its role as a transcriptional regulator emerged from genetic and molecular studies in which its homology with CRP (the cyclic AMP receptor protein which mediates catabolite repression) was established and has since been particularly important in identifying the structural basis of its regulatory specificities. FNR is a member of a growing family of CRP-related regulatory proteins which have a DNA-binding domain based on the helix-turn-helix structural motif, and a characteristic beta-roll that is involved in nucleotide-binding in CRP. The FNR protein has been isolated in a monomeric form (Mr 30,000) which exhibits a high but as yet non-specific affinity for DNA. Nevertheless, the DNA-recognition site and important residues conferring the functional specificity of FNR have been defined by site-directed mutagenesis. A consensus for the sequences that are recognized by FNR in the promoter regions of FNR-regulated genes, has likewise been identified. The basic features of the genes and operons regulated by FNR are reviewed, and examples in which FNR functions negatively as an anaerobic repressor as well as positively as an anaerobic activator, are included. Less is known about the way in which FNR senses anoxia and is thereby transformed into its 'active' form, but it seems likely that cysteine residues and possibly a metal ion are involved. Four of the five cysteine residues of FNR are clustered in an essential N-terminal 'domain' which is conserved in FNR and the HlyX protein of Actinobacillus pleuropneumoniae, but not in CRP or the FixK protein of Rhizobium meliloti. The relationships between FNR and other oxygen-related systems in E. coli are discussed, as well as parallel systems in other organisms.(ABSTRACT TRUNCATED AT 400 WORDS)