Résumé : No large study to date has ever evaluated the effectiveness, safety and tolerability of imipenem/clavulanate versus meropenem/clavulanate to treat multidrug- and extensively drug-resistant tuberculosis (MDR- and XDR-TB). The aim of this observational study was to compare the therapeutic contribution of imipenem/clavulanate versus meropenem/clavulanate added to background regimens to treat MDR- and XDR-TB cases. 84 patients treated with imipenem/clavulanate-containing regimens showed a similar median number of antibiotic resistances (8 versus 8) but more fluoroquinolone resistance (79.0% versus 48.9%, p<0.0001) and higher XDR-TB prevalence (67.9% versus 49.0%, p=0.01) in comparison with 96 patients exposed to meropenem/clavulanate-containing regimens. Patients were treated with imipenem/clavulanate- and meropenem/clavulanate-containing regimens for a median (interquartile range) of 187 (60-428) versus 85 (49-156) days, respectively. Statistically significant differences were observed on sputum smear and culture conversion rates (79.7% versus 94.8%, p=0.02 and 71.9% versus 94.8%, p<0.0001, respectively) and on success rates (59.7% versus 77.5%, p=0.03). Adverse events to imipenem/clavulanate and meropenem/clavulanate were reported in 5.4% and 6.5% of cases only. Our study suggests that meropenem/clavulanate is more effective than imipenem/clavulanate in treating MDR/XDR-TB patients.