INTRODUCTION: There is limited literature on the appropriateness of viral load (VL) monitoring and management of detectable VL in public health settings in rural South Africa. METHODS: We analysed data captured in the electronic patient register from HIV-positive patients >/= 15 years old initiating antiretroviral therapy (ART) in 17 public sector clinics in rural KwaZulu-Natal, during 2010-2016. We estimated the completion rate for VL monitoring at 6, 12, and 24 months. We described the cascade of care for those with any VL measurement >/= 1000 HIV-1 RNA copies/mL after >/= 20 weeks on ART, including the following proportions: (1) repeat VL within 6 months; (2) re-suppressed; (3) switched to second-line regimen. RESULTS: There were 29 384 individuals who initiated ART during the period [69?male, median age 31 years (interquartile range 25-39)]. Of those in care at 6, 12, and 24 months, 40.7% (9861/24 199), 34% (7765/22 807), and 25.5% (4334/16 965) had a VL test at each recommended time-point, respectively. The VL results were documented at all recommended time-points for 12% (2730/22 807) and 6.2% (1054/16 965) of ART-treated patients for 12 and 24 months, respectively. Only 391 (18.3%) of 2135 individuals with VL >/= 1000 copies/mL on first-line ART had a repeat VL documenting re-suppression or were appropriately changed to second-line with persistent failure. Completion of the treatment failure cascade occurred a median of 338 days after failure was detected. CONCLUSION: We found suboptimal VL monitoring and poor responses to virologic failure in public-sector ART clinics in rural South Arica. Implications include increased likelihood of morbidity and transmission of drug-resistant HIV.