OBJECTIVE: HIV viral load (VL) monitoring is generally conducted 6-12 monthly in low- and middle-income countries, risking relatively prolonged periods of poor viral control. We explored the effects of different levels of loss of viral control on immune reconstitution and activation. DESIGN: : 208 participants starting Protease Inhibitor (PI)-based second-line therapy in the EARNEST trial (ISRCTN37737787) in Uganda and Zimbabwe were enrolled and CD38+/HLA-DR+ immunophenotyping performed (CD8-FITC/CD38-PE/CD3-PerCP/HLA-DR-APC; centrally gated) in real-time at 0, 12, 48, 96 and 144 weeks from randomisation. METHODS: Viral Load (Viral load (VL) was assayed retrospectively on samples collected every 12-16 weeks and classified as (1) continuous suppression (<40 copies/ml throughout); (2) suppression with transient blips; (3) low-level rebound (two or more consecutive VL >40, <5000 copies/ml); (4) high-level rebound/non-response (two or more consecutive VL >5000 copies/ml). RESULTS: Immunophenotype reconstitution varied between that defined by numbers of cells and that defined by cell percentages. Furthermore, VL dynamics were associated with substantial differences in expression of CD4+ and CD8+ cell activation markers, with only individuals with high-level rebound/non-response (>5000 copies/ml) experiencing significantly greater activation and impaired reconstitution. There was little difference between participants who suppressed consistently and who exhibited transient blips or even low-level rebound by 144 weeks (p > 0.2 vs suppressed consistently). CONCLUSION: Detectable viral load below the threshold at which WHO guidelines recommend that treatment can be maintained without switching (1000 copies/ml) appear to have at most, small effects on reconstitution and activation, for patients taking a PI-based second-line regimen.