BACKGROUND: Xpert® MTB/RIF was expected to revolutionise the management of rifampicin-resistant TB (RR-TB) by enabling rapid and decentralised diagnosis of rifampicin (RIF) resistance.METHODS: We performed a care cascade analysis for a cohort of RR-TB patients managed under programmatic conditions. Cumulative incidences of time to completion of the RR-TB care cascade steps were estimated, reasons for delay or attrition from the cascade investigated and WHO programme indicators for monitoring of RR-TB programmes calculated.RESULTS: Of 502 patients diagnosed with RR-TB using Xpert, 64% initiated multidrug-resistant TB (MDR-TB) treatment immediately, 20% after some first-line treatment, 16% never initiated MDR-TB treatment, mainly because of death (44%) or loss to follow-up (26%) soon after diagnosis. A supplementary sputum sample was collected within 14 days of treatment in 58.8% of cases. Only 63% of RR-TB cases were assessed for isoniazid resistance, and only 65% of MDR-TB cases were evaluated for pre-XDR-TB (extensively drug-resistant TB). Treatment was individualised in 57% of pre-XDR and 68% of XDR-TB patients. Only 8% completed the entire RR-TB care cascade as intended.CONCLUSION: Fidelity to the RR-TB algorithm was poor, with substantial losses at each step of the cascade, highlighting the fact that implementation of novel technologies needs to be accompanied by health system strengthening to maximise impact.