Accurate identification of bacterial isolates is an essential task in clinical microbiology. Phenotypic methods are time-consuming and either fail to identify some bacteria such as Gram-positive rods entirely or at least fail to do so in some clinical situations. 16S rDNA sequencing is a recent method of identification which offers a useful alternative. In this study, we investigate the usefulness of this method for identifying a range of bacteria in a clinical laboratory under routine conditions. Over a period of 30 months, 683 isolates were obtained from clinical specimens, sequenced and analysed. For 568 of these isolates (83.1%), the sequence provided species level identification. For 108 isolates (15.8%), the identification was limited to the genus level, and for 7 isolates (1%), the sequence remained unidentifiable by 16S rDNA sequence analysis. For the isolates identified only to the genus level, the 16S rDNA approach failed to identify bacteria to the taxonomic level for 3 reasons: failure to differentiate between species in 72 isolates (66%), the lack of any closely related sequence in the database for 15 isolates (13.8%) and the presence of more than 1% of undetermined position in the sequence for 13 isolates (12%).