As the prevalence of multidrug-resistant and extensively drug-resistant tuberculosis strains continues to rise, so does the need to develop accurate and rapid molecular tests to complement time-consuming growth-based drug susceptibility testing. Performance of molecular methods relies on the association of specific mutations with phenotypic drug resistance and while considerable progress has been made for resistance detection of first-line antituberculosis drugs, rapid detection of resistance for second-line drugs lags behind. The rrs A1401G allele is considered a strong predictor of cross-resistance between the three second-line injectable drugs, capreomycin (CAP), kanamycin, and amikacin. However, discordance is often observed between the rrs A1401G mutation and CAP resistance, with up to 40% of rrs A1401G mutants being classified as CAP susceptible. We measured the MICs to CAP in 53 clinical isolates harboring the rrs A1401G mutation and found that the CAP MICs ranged from 8 μg/ml to 40 μg/ml. These results were drastically different from engineered A1401G mutants generated in isogenic Mycobacterium tuberculosis, which exclusively exhibited high-level CAP MICs of 40 μg/ml. These data support the results of prior studies, which suggest that the critical concentration of CAP (10 μg/ml) used to determine resistance by indirect agar proportion may be too high to detect all CAP-resistant strains and suggest that a larger percentage of resistant isolates could be identified by lowering the critical concentration. These data also suggest that differences in resistance levels among clinical isolates are possibly due to second site or compensatory mutations located elsewhere in the genome.