Inflammation influences the assessment of nutritional status. For example, inflammation reduces plasma retinol concentrations and vitamin A deficiency is overestimated. Conversely inflammation increases plasma ferritin concentrations and Fe deficiency is underestimated. Blood samples were obtained from 163 free-living HIV-1-infected adults, not on continuous medication, anti-retroviral drugs or micronutrients, not unwell and who had not reached WHO stage IV of HIV/AIDS. We used four markers of inflammation, C-reactive protein (CRP), α1-acid glycoprotein (AGP), α1-antichymotrypsin and erythrocyte sedimentation rate but mainly CRP and AGP were used to separate the subjects into four groups: ‘healthy’ where both CRP and AGP were normal; ‘incubation phase’ where CRP was elevated; ‘early convalescence’ where AGP and CRP were elevated and ‘late convalescence’ where only AGP was elevated. Correction factors were calculated to remove the influence of inflammation from each biomarker and group where inflammation was present and the data are shown before and after recalculation. The correction increased median plasma retinol concentrations of the whole group from 1·16 to 1·33 μmol/l, comparable with values (mean 1·29 μmol/l) in HIV-negative Kenyan women. Median ferritin concentrations fell by about 50 % in both sexes and the number of women with plasma ferritin concentrations ≤ 12 μg/l increased from eleven to twenty. The correction also increased plasma carotenoids and Hb but not α-tocopherol concentrations. We suggest that the method described to remove the influence of inflammation from nutritional biomarkers should be generally applicable in apparently healthy people and prevents discarding valuable data because of mild inflammation. The method does now need to be tested in other populations.