Résumé : Background: Improvements in cognitive function are described after initiation of combination antiretroviral therapy (cART), with sparse data on differences between cART strategies. Methods: We assessed changes in cognition, over 96 weeks, in therapy-naive HIV-positive adults randomized to darunavir/ ritonavir (800/100 mg once daily) with either raltegravir (400 mg twice daily, Arm1) or tenofovir/emtricitabine (245/200 mg once daily, Arm2). Seven cognitive tests were administered at baseline and week (W) 96. Changes from baseline in individual cognitive test scores and composite score (NPZ) were assessed. Comparisons between treatment arms were by intention to treat and associations with immunological and virological parameters by regression models. Findings: Of 343 subjects enrolled, 208 completed the W96 cognitive assessment. Baseline median (interquartile range) CD4+ count and plasma HIV RNA were 348 (282-398) cells per microliter and 4.7 (4.2-5.1) log10 copies per milliliter, respectively. At W96, numbers with plasma HIV RNA undetectable and remaining on randomized cART were 85 (92%) and 110 (96%), and 84 (90%) and 107 (93%) in Arm1 and Arm2, respectively. Overall performance significantly improved by W96 in 5 of 7 individual tests and in NPZ. Mean changes in NPZ were 0.28 versus 0.21 for Arm1 and 2, respectively (P = 0.37). No statistically significant differences between study treatment arms were observed in individual cognitive domains apart from attention (greater improvement in Arm1, P = 0.0499). At W96, NPZ score increase was associated with increase in CD4+ (P = 0.001) but not HIV RNA area under curve (P = 0.60). Interpretation: Subsequent to the initiation of cART, immunological recovery rather than type of antiretroviral therapy is the major driver of changes in cognitive function.