Bone and joint infections (BJI) are severe infections that require a tailored and protracted antibiotic treatment. Yet, the diagnostic based on culturing samples lacks sensitivity, especially for hardly culturable bacteria. Metagenomic sequencing could potentially address those limitations. Here, we assessed the performances of metagenomic sequencing on 24 BJI samples for the identification of pathogens and the prediction of their antibiotic susceptibility. For monomicrobial samples in culture (n = 8), the presence of the pathogen was confirmed by metagenomics in all cases. For polymicrobial samples (n = 16), 32/55 bacteria (58.2%) were found at the species level (and 41/55 [74.5%] at the genus level). Conversely, 273 bacteria not found in culture were identified, 182 being possible pathogens and 91 contaminants. A correct antibiotic susceptibility could be inferred in 94.1% and 76.5% cases for monomicrobial and polymicrobial samples, respectively. Altogether, we found that clinical metagenomics applied to BJI samples is a potential tool to support conventional culture. Context. Bone and joint infections (BJI) are severe infections that affect a growing number of patients 1. Along with the surgical intervention, the microbiological diagnosis is a keystone of the management of BJI in (i) identifying the bacteria causing the infection and (ii) assessing their susceptibility to antibiotics. Currently, this is achieved by culturing surgical samples on various media and conditions, together with a long time of incubation to recover fastidiously-growing bacteria that can be involved in BJI. Still, some bacteria will not grow under these conditions because of extreme oxygen sensitivity, a prior antibiotic intake or metabolic issues (e.g. quiescent bacteria in chronic infections). Consequently, the antibiotic treatment may not span all the bacteria involved in the infection, which can favour the relapse and the need for a new surgery. Clinical metagenomics refers to the concept of sequencing all the DNA (i.e. all the genomes) present in a clinical sample with the purpose of identifying pathogens and inferring their antibiotic susceptibility pattern 2. This new, culture-independent method takes advantages of the thrilling development of next-generation sequenc-ing (NGS) technologies since the mid-2000s. The NGS platforms typically yield thousands to millions of reads (sequences of size ranging from 100 bp to a few kbp), which virtually enables to recover the sequences of all the genes present in the sample, yet in a disorganised fashion. Substantial bio-informatics efforts are thereby needed to reconstruct and reorder the original sequences in genomes, and are referred to as the assembly process. Hence, various information such as the taxonomic identification of the present species, antibiotic resistance determinants (ARDs), mutations (as compared to a reference genome or sequence), single nucleotide variants (SNVs, for clonality assessment) and virulence genes can be determined. Clinical metagenomics is an emerging field in medicine. So far, a few attempts to use metagenomics on clinical samples have been performed (on urines 3, 4 , cerebrospinal fluid or brain biopsy 5, 6 , blood 7 and skin granuloma 8) likely because of the high price of metagenomics and the complexity of the management of sequence data for clinical microbiologists. To the best of our knowledge, metagenomics has never been applied to BJI samples.