OBJECTIVES: To characterize rifampicin-resistant strains missed by the Mycobacteria Growth Indicator Tube (MGIT) 960 system but not by egg-based media in the UK and Ireland and to ascertain their prevalence. METHODS: All strains sent for second-line susceptibility testing were prospectively collected. Drug Susceptibility Testing was performed by Resistance Ratio (RR), Proportion Method (PM), MGIT 960 and MIC determination by microdilution. Rifampicin-resistance-conferring mutations were detected with line probe assays and sequencing. At the end of the study period, retrospective archived strains from 2010 to 2014 showing key mutations were analysed phenotypically and genotypically. RESULTS: Seventeen of 7234 prospective isolates were included. All of them were susceptible by MGIT. One was borderline by RR (MIC to rifampicin of 4 mg/L) and was resistant by PM. Eight were resistant and eight were highly resistant on RR. These 16 isolates had MICs between 1 and 8 mg/L on microdilution. With PM, 16/17 were susceptible to rifampicin. 17/17 had mutations in the rpoB gene. D516Y was the mutation most frequently found (13/17). Retrospectively, ten additional strains with key genotypes were found in our collection: 6/10 were susceptible in the MGIT and resistant in RR. Of the 27 studied strains, the MGIT only detected resistance in four. CONCLUSIONS: Rifampicin resistance is missed by the MGIT system. In the UK and Ireland the prevalence of these strains is low. The introduction of routine molecular testing would detect false susceptibility. Further research is needed to ascertain the role of these strains in clinical failure and their prevalence in other settings.