Until recently it was believed that most Bacteria form Gln-tRNA(GLN) by the amidation of Glu-tRNA(GLN), only a few members of the gamma subdivision of Proteobacteria being able to charge tRNA(GLN) directly. We undertook a phylogenetic study in an attempt to determine at what point the changeover to the direct system may have occurred. To this end, we selected a number of representative Proteobacteria to see if we could find a division point. We constructed degenerate primers and conducted PCR analysis to identify which Bacteria had Gln-tRNA synthetase, on the one hand, and which had the amidotransferase system, on the other. At the same time, we surveyed data banks of completely sequenced microbial genomes, as well as those for genomes in the process of being sequenced. These combined efforts revealed four Proteobacteria in a phylogenetically intermediate position which have the genetic potential for both mechanisms. Perplexingly, however, three distantly related bacteria were also found to have both enzymes.