The Burkholderia pseudomallei multilocus sequence typing (MLST) database (http://pubmlst.org/bpseudomallei/) contains the largest global sequence repository for B. pseudomallei and its closest genetic relatives. Using conventional MLST and in silico MLST data derived from publicly available whole-genome sequences, we first defined the phylogenetic relatedness of B. pseudomallei and its nearest neighbours. Based on this analysis, we propose that the recently described B. pseudomallei complex (Bpc) should be expanded to encompass B. pseudomallei, Burkholderia humptydooensis (proposed), Burkholderia mallei, Burkholderia oklahomensis, Burkholderia thailandensis and three unassigned Burkholderia Clades A, B and C (represented by type strains BDU 5, BDU 8 and MSMB0265, respectively). Of note, the MLST narK locus is present in all Bpc species but is missing in all other Burkholderia spp., including all Burkholderia cepacia complex species, with the exception of most Burkholderia ubonensis strains, which contain narK but encode genetically distinct sequences. The presence of narK is thus indicative of a Bpc strain. Next, we revisited in silico the performance of the existing MLST primers, which prompted redesign of primers targeting the gmhD, lepA, lipA, narK and ndh loci to encompass genetic diversity among Bpc strains and to address amplification/sequencing issues. We show in silico and in vitro that the redesigned primers yield good-quality amplification and sequencing results for the gmhD, lepA, lipA, narK and ndh loci in Bpc species. These primers provide an alternative for amplification and sequencing of MLST loci in Bpc species in cases when poor-quality amplification or sequencing data are obtained using the original MLST primers.