It has long been assumed that some rifampicin-resistant Mycobacterium tuberculosis strains are susceptible to, and thus treatable with, rifabutin. However, clinical breakpoints for susceptibility testing of rifabutin as well as the evidence for a clinical effect of rifabutin in rifampicin-resistant strains remains poorly defined. The objective of this study was to re-evaluate the breakpoint for rifabutin in relation to its MIC wild-type distribution and the presence of mutations in rpoB. The MIC in 7H10 Middlebrook medium was determined for clinical isolates of M. tuberculosis (n = 95), where a majority were multidrug resistant. Additionally, all strains were screened for rpoB mutations by sequencing and the GenoType MTBDRplus assay. Rifampicin resistance was confirmed by genotypical and/or phenotypical tests in 73 isolates (76.8%). Nineteen isolates, defined as rifampicin resistant and rifabutin susceptible according to the present breakpoint, exhibited significantly higher MICs of rifabutin (0.064-0.5 mg/L) than rifabutin-susceptible isolates without any detectable mutations in rpoB (P < 0.001). These 19 isolates were clearly resistant to rifampicin (MIC 2-256 mg/L) and all but one had mutations in rpoB, with 9 (47.4%) specifically in Asp516Val. Our results indicate that rifampicin-resistant but rifabutin-susceptible isolates according to the present breakpoints harbour rpoB mutations and have a rifabutin MIC significantly higher than strains without any detectable mutations in rpoB. So far there are no clinical, pharmacological or microbiological data to confirm that such isolates can be treated with rifabutin and we suggest a revision of the current breakpoints.