Pulmonary tuberculosis is a leading cause of morbidity and mortality in developing countries. Drug resistance, a huge problem in this contagious disease, is driven by point mutations in the Mycobacterium tuberculosis genome however, their frequencies vary geographically and this affects applicability of molecular diagnostics for rapid detection of resistance. Here, we report the frequency and patterns of mutations associated with resistance to second-line anti-TB drugs in multidrug-resistant (MDR) M. tuberculosis isolates from eSwatini, Somalia and Uganda that were resistant to a second-line anti-TB drug.The quinolone resistance determining region (QRDR) of gyrA/gyrB genes and the drug resistance associated fragment of rrs gene from 80 isolates were sequenced and investigated for presence of drug resistance mutations. Of the 80 isolates, 40 were MDR, of which 28 (70%) were resistant to a second-line anti-TB injectable drug, 18 (45%) were levofloxacin resistant while 12 (30%) were extensively drug resistant (XDR). The remaining 40 isolates were susceptible to anti-TB drugs. MIRU-VNTR analysis was performed for M/XDR isolates.We successfully sub-cultured 38 of the 40 M/XDR isolates. The gyrA resistance mutations (Gly88Ala/Cys/Ala, Ala90Val, Ser91Pro, Asp94Gly/Asn) and gyrB resistance mutations (Asp500His, Asn538Asp) were detected in 72.2% (13/18) and 22.2% (4/18) of the MDR and levofloxacin resistant isolates, respectively. Overall, drug resistance mutations in gyrA/gyrB QRDRs occurred in 77.8% (14/18) of the MDR and levofloxacin resistant isolates. Furthermore, drug resistance mutations a1401g and g1484 t in rrs occurred in 64.3% (18/28) of the MDR isolates resistant to a second-line anti-TB injectable drug. Drug resistance mutations were not detected in drug susceptible isolates.The frequency of resistance mutations to second-line anti-TB drugs in MDR-TB isolates resistant to second line anti-TB drugs from eSwatini, Somalia and Uganda is high, implying that rapid molecular tests are useful in detecting second-line anti-TB drug resistance in those countries. Relatedly, the frequency of fluoroquinolone resistance mutations in gyrB/QRDR is high relative to global estimates, and they occurred independently of gyrA/QRDR mutations implying that their absence in panels of molecular tests for detecting fluoroquinolone resistance may yield false negative results in our setting.