Background: Lack of viral load monitoring of ART is known to be associated with slower switch from a failing regimen and thereby higher prevalence of MDR HIV-1. Many countries have continued to use thymidine analogue drugs despite recommendations to use tenofovir in combination with a cytosine analogue and NNRTI as first-line ART. The effect of accumulated thymidine analogue mutations (TAMs) on phenotypic resistance over time has been poorly characterized in the African setting. Patients and methods: A retrospective analysis of individuals with ongoing viral failure between weeks 48 and 96 in the NORA (Nevirapine OR Abacavir) study was conducted. We analysed 36 genotype pairs from weeks 48 and 96 of first-line ART (14 treated with zidovudine/lamivudine/nevirapine and 22 treated with zidovudine/lamivudine/abacavir). Phenotypic drug resistance was assessed using the Antivirogram assay (v. 2.5.01, Janssen Diagnostics). Results: At 96 weeks, extensive TAMs (>/=3 mutations) were present in 50% and 73% of nevirapine- and abacavir-treated patients, respectively. The mean (SE) number of TAMs accumulating between week 48 and week 96 was 1.50 (0.37) in nevirapine-treated participants and 1.82 (0.26) in abacavir-treated participants. Overall, zidovudine susceptibility of viruses was reduced between week 48 [geometric mean fold change (FC) 1.3] and week 96 (3.4, P = 0.01). There was a small reduction in tenofovir susceptibility (FC 0.7 and 1.0, respectively, P = 0.18). Conclusions: Ongoing viral failure with zidovudine-containing first-line ART is associated with rapidly increasing drug resistance that could be mitigated with effective viral load monitoring.