Abstract Objectives The dolutegravir/valproic acid drug–drug interaction (DDI) is suggested to be caused by protein displacement. Here, we assess the underlying mechanism. Methods Participants in a randomized controlled trial investigating valproic acid as an HIV latency reversing agent were recruited in a predefined pharmacokinetic substudy if they were on once-daily 50 mg dolutegravir-containing combination ART (cART) for >12 months with a plasma HIV-RNA <50 copies/mL (trial registration: ClinicalTrials.gov NCT03525730). Participants were randomized to receive 30 mg/kg/day valproic acid orally (divided into two equal doses) for 14 days or not. Total and unbound dolutegravir concentrations were measured on day 0 (before intake of valproic acid and 6 h after intake of valproic acid) and on days 1, 7, 14 and 42. Intra- and inter-subject dolutegravir concentrations and geometric means (GMs) were evaluated. Results Six of 10 participants on dolutegravir were randomized to receive valproic acid. During 14 days of valproic acid treatment, the GM total dolutegravir concentration decreased sharply from 1.36 mg/L on day 0 to 0.85, 0.31 and 0.20 mg/L on days 0, 1, 7 and 14, respectively, while total dolutegravir concentrations in the controls remained comparable during the same period: 1.27–1.49 mg/L. We observed a parallel increase in unbound dolutegravir fractions ranging from 0.39% to 0.58% during valproic acid administration, compared with 0.25% to 0.28% without valproic acid. Unbound dolutegravir concentrations were above the established in vitro EC90 value for unbound dolutegravir in 85% of the tested samples. Conclusions This study confirms protein displacement as the main mechanism for this DDI, although additional mechanisms might be involved too. If dolutegravir is taken with food, this DDI is probably not clinically relevant.