BACKGROUND: People with HIV accessing antiretroviral therapy (ART) have persistent physical, psychological, social, and spiritual problems, which are associated with poor quality of life and treatment outcomes. We assessed the effectiveness of a nurse-led palliative care intervention on patient-reported outcomes. METHODS: We did this randomised controlled trial at a clinic in Kenya for adults with HIV, established on ART, and reporting moderate-to-severe pain or symptoms. We randomly assigned participants (1:1) either to a palliative care intervention (including assessments of physical, emotional, and spiritual wellbeing and quality of life) given six times over 4 months, or to usual care. Participants and investigators were not masked to allocation. The primary outcome was pain (scored on the African Palliative Care Association's African Palliative Outcome Scale). This trial is registered with ClinicalTrials.gov, number NCT01608802. FINDINGS: We screened 2070 patients, of whom we enrolled 120: 60 allocated to each group. In the control group, median pain score improved from 1·0 (IQR 0·0-2·0) at baseline to 5·0 (3·0-5·0) at 4 months; in the intervention group, it improved from 1·0 (0·0-2·0) at baseline to 4·5 (3·0-5·0) at 4 months. Compared with standard care, the intervention had no significant effect on pain (coefficient -0·01, 95% CI -0·36 to 0·34, p=0·95). INTERPRETATION: A nurse-led palliative care intervention was not effective in reducing pain. However, person-centred assessment and care delivered by staff who have received additional training had positive effects on self-reported mental health related quality of life and psychosocial wellbeing. FUNDING: Diana Princess of Wales Memorial Fund.