Objectives Emerging resistance to antiretroviral drugs may jeopardize the achievements of improved access to ART. We compared the prevalence of different resistance mutations in HIV-infected adults with virological failure in a cohort with regular routine viral load (VL) monitoring (Switzerland) and cohorts with limited access to VL testing (Uganda and Lesotho). Methods We considered individuals who had genotypic resistance testing (GRT) upon virological failure (≥1000 copies/mL) and were on ART consisting of at least one NNRTI and two NRTIs. From Lesotho, individuals with two subsequent VLs ≥1000 copies/mL despite enhanced adherence counselling (n = 58) were included in the analysis. From Uganda, individuals with a single VL ≥1000 copies/mL (n = 120) were included in the analysis. From the Swiss HIV Cohort Study (SHCS), a population without history of monotherapy or dual therapy with the first GRT upon virological failure (n = 61) was selected. Results We found that 50.8% of individuals in the SHCS, 72.5% in Uganda and 81.0% in Lesotho harboured HIV with high-level resistance to at least two drugs from their current regimen. Stanford resistance scores were higher in Uganda compared with Switzerland for all drugs used in first-line treatment except zidovudine and tenofovir (P < 0.01) and higher in Lesotho compared with Uganda for all drugs used in first-line treatment except zidovudine (P < 0.01). Conclusions Frequent VL monitoring and possibly pretreatment GRT as done in the SHCS pays off by low levels of resistance even when treatment failure occurs. The high-level resistance patterns in Lesotho compared with Uganda could reflect a selection of strains with multiple resistance during enhanced adherence counselling.