ObjectivesHeteroresistant infections are defined as infections in which a mixture of drug-resistant and drug-susceptible populations are present. InMycobacterium tuberculosis(M. tb), heteroresistance poses a challenge in diagnosis and has been linked with poor treatment outcomes. We compared the analytical sensitivity of molecular methods, such as GeneXpert and whole genome sequencing (WGS) in detecting heteroresistance when compared with the ‘gold standard’ phenotypic assay: the agar proportion method (APM).MethodsUsing two rounds of proficiency surveys with defined monoresistant BCG strains and mixtures of susceptible/resistantM. tb, we determined the limit of detection (LOD) of known resistance associated mutations.ResultsThe LOD for rifampin-R (RIF-R) detection was 1% using APM, 60% using GeneXpert MTB/RIF, 10% using GeneXpert MTB/RIF Ultra and 10% using WGS. While WGS could detect mutations beyond those associated with RIF resistance, the LOD for these other mutations was also 10%. Additionally, we observed instances where laboratories did not report resistance in the majority population, yet the mutations were present in the raw sequence data.ConclusionThe gold standard APM detects minority resistant populations at a lower proportion than molecular tests.Mycobacterium bovisBCG strains with defined resistance and extracted DNA fromM. tbprovided concordant results and can serve in quality control of laboratories offering molecular testing for resistance. Further research is required to determine whether the higher LOD of molecular tests is associated with negative treatment outcomes.