Background Since a meta-analysis showed little or no effect of second-line injectables on treatment success, and using injectables may induce ototoxicity, injectable-free rifampicin-resistant tuberculosis (RR-TB) treatment regimens are recommended. However, acquired resistance preventing activity was overlooked. No previous study assessed the effect of shortening the duration of kanamycin administration to 2 months during the intensive phase of the RR-TB shorter treatment regimen (STR). Methods Retrospective cohort study of the effect of using 2 months of kanamycin instead of the standard 4(+) months (extension if lack of smear conversion at 4 months) on recurrence (either treatment failure or relapse) and fluoroquinolone acquired drug resistance, in patients treated with a gatifloxacin-based STR in Damien Foundation supported clinics in Bangladesh. Logistic regression was used to estimate associations. Results Five of 52 (9.6%) treated with a STR containing two months of kanamycin had recurrence, compared to 21 of 738 (2.8%) patients treated with the standard STR containing 4(+) months of kanamycin (OR 3.7; 95%CI:1.5-10.3). In those with initially fluoroquinolone-susceptible TB, acquired resistance to fluoroquinolone was detected in none of 639 patients treated with 4(+) months of kanamycin and two (4.5%) of 44 treated with two months of kanamycin (OR 75.2; 95%CI:3.6-1592.1). Conclusion Two months of kanamycin was insufficient to prevent recurrence with acquired resistance to gatifloxacin, the core drug of the most effective RR-TB STR. Injectable mediated resistance