OBJECTIVE: To deliver rapid isoniazid (INH) and rifampicin (RMP) drug susceptibility testing (DST) close to the patient, we designed a decentralisation process for the microscopic observation drug susceptibility (MODS) assay in Peru and evaluated its reliability. METHODS: After 2 weeks of training, laboratory staff processed ≥120 consecutive sputum samples each in three regional laboratories. Samples were processed in parallel with MODS testing at an expert laboratory. Blinded paired results were independently analysed by the Instituto Nacional de Salud (INS) according to predetermined criteria: concordance for culture, DST against INH and RMP and diagnosis of multidrug-resistant tuberculosis (MDR-TB) ≥ 95%, McNemar's P > 0.05, kappa index (κ) ≥ 0.75 and contamination 1-4%. Sensitivity and specificity for MDR-TB were calculated. RESULTS: The accreditation process for Callao (126 samples, 79.4% smear-positive), Lima Sur (n = 130, 84%) and Arequipa (n = 126, 80%) took respectively 94, 97 and 173 days. Pre-determined criteria in all regional laboratories were above expected values. The sensitivity and specificity for detecting MDR-TB in regional laboratories were <95%, except for sensitivity in Lima Sur, which was 91.7%. Contamination was 1.0- 2.3%. Mean delay to positive MODS results was 9.9-12.9 days. CONCLUSION: Technology transfer of MODS was reliable, effective and fast, enabling the INS to accredit regional laboratories swiftly.