The mycobacterial cell wall core consists of an outer lipid layer of mycolic acids connected, via arabinogalactan polysaccharide, to an inner peptidoglycan layer. An alpha-L-rhamnopyranosyl residue has been shown to be a key component linking the mycolated arabinogalactan to the peptidoglycan and, therefore, the biosynthesis of L-rhamnose (Rha) in mycobacteria was investigated as the first step of developing inhibitors of its biosynthesis. Biochemical assays were used to show that dTDP-Rha was synthesized in Mycobacterium smegmatis from alpha-D-glucose 1-phosphate (alpha-D-Glc-1-P) and dTTP by the same four enzymic steps used by Escherichia coli and other bacteria. PCR primers based on consensus regions of known sequences of the first enzyme in this series, alpha-D-Glc-1-P thymidylytransferase (RfbA) were used to amplify rfbA DNA from M. tuberculosis. The entire rfbA gene was then cloned and sequenced. The deduced amino acid sequence revealed a 31362 Da putative protein product which showed similarity to RfbA proteins of other bacteria (59% identity to that found in E. coli). Sequencing of DNA flanking the rfbA gene did not reveal any of the other rfb genes required for dTDP-Rha biosynthesis. Therefore, the four Rha biosynthetic genes are not clustered in M. tuberculosis. The enzymic activity of the sequenced gene product was confirmed by transformation of E. coli with pBluescript KS(-) containing the rfbA gene from M. tuberculosis. Analysis of enzyme extracts prepared from this transformant revealed an 11-fold increase in alpha-D-Glc-1-P thymidylyltransferase activity.