<jats:title>Abstract</jats:title><jats:sec id="sa1"><jats:title>Background</jats:title><jats:p><jats:italic>Klebsiella pneumoniae</jats:italic>and close relatives are a growing cause of healthcare-associated infections for which increasing rates of multi-drug resistance are a major concern. The<jats:italic>Klebsiella</jats:italic>polysaccharide capsule is a major virulence determinant and epidemiological marker. However, little is known about capsule epidemiology since serological typing is not widely accessible, and many isolates are serologically non-typeable. Molecular methods for capsular typing are needed, but existing methods lack sensitivity and specificity and fail to take advantage of the information available in whole-genome sequence data, which is increasingly being generated for surveillance and investigation of<jats:italic>Klebsiella</jats:italic>.</jats:p></jats:sec><jats:sec id="sa2"><jats:title>Methods</jats:title><jats:p>We investigated the diversity of capsule synthesis loci (K loci) among a large, diverse collection of 2503 genome sequences of<jats:italic>K. pneumoniae</jats:italic>and closely related species. We incorporated analyses of both full-length K locus DNA sequences and clustered protein coding sequences to identify, annotate and compare K locus structures, and we propose a novel method for identifying K loci based on full locus information extracted from whole genome sequences.</jats:p></jats:sec><jats:sec id="sa3"><jats:title>Results</jats:title><jats:p>A total of 134 distinct K loci were identified, including 31 novel types. Comparative analysis of K locus gene content detected 508 unique protein coding gene clusters that appear to reassort via homologous recombination, generating novel K locus types. Extensive nucleotide diversity was detected among the<jats:italic>wzi</jats:italic>and<jats:italic>wzc</jats:italic>genes, both within and between K loci, indicating that current typing schemes based on these genes are inadequate. As a solution, we introduce<jats:italic>Kaptive</jats:italic>, a novel software tool that automates the process of identifying K loci from large sets of<jats:italic>Klebsiella</jats:italic>genomes based on full locus information.</jats:p></jats:sec><jats:sec id="sa4"><jats:title>Conclusions</jats:title><jats:p>This work highlights the extensive diversity of<jats:italic>Klebsiella</jats:italic>K loci and the proteins that they encode. We propose a standardised K locus nomenclature for<jats:italic>Klebsiella</jats:italic>, present a curated reference database of all known K loci, and introduce a tool for identifying K loci from genome data (<jats:ext-link xmlns:xlink="http://www.w3.org/1999/xlink" ext-link-type="uri" xlink:href="https://github.com/katholt/Kaptive">https://github.com/katholt/Kaptive</jats:ext-link>). These developments constitute important new resources for the<jats:italic>Klebsiella</jats:italic>community for use in genomic surveillance and epidemiology.</jats:p></jats:sec>