Background: Paediatric data on CNS penetration of antiretroviral drugs are scarce. Objectives: To evaluate CNS penetration of antiretroviral drugs in HIV-infected children and explore associations with neurocognitive function. Patients and methods: Antiretroviral drug levels were measured in paired CSF and blood samples of clinically stable HIV-infected children between 8 and 18 years old on long-term combined ART. Plasma drug concentrations were corrected for protein binding. We evaluated CNS penetration using CSF/plasma ratios and compared CSF concentrations with the IC50 as a surrogate marker for effectiveness. Blood-brain barrier permeability was assessed for possible confounding. Associations with neurocognitive function were explored using linear regression analysis. Results: Median CSF/plasma ratios (IQR) were: lopinavir 0.059 (0.024-0.157, n = 7), efavirenz 0.681 (0.555-0.819, n = 12), tenofovir 0.021 (0.020-0.024, n = 4), lamivudine 0.464 (0.331-0.607, n = 17), emtricitabine 0.365 (0.343-0.435, n = 3), nevirapine 1.203 (n = 1), zidovudine 0.718 (0.711-1.227, n = 5) and abacavir 1.344 (0.670-2.450, n = 10). CSF concentrations were below the IC50 for tenofovir (100%), emtricitabine (100%), abacavir (50%) and zidovudine (17%). Lamivudine, lopinavir, efavirenz and nevirapine concentrations were all above the IC50. All participants were virologically suppressed in blood and CSF. CSF drug concentrations were not associated with blood-brain barrier permeability or neurocognitive function. Conclusions: We showed adequate CSF concentrations of lamivudine, lopinavir, efavirenz and nevirapine, and potential suboptimal CSF concentrations of tenofovir, abacavir and emtricitabine in long-term treated HIV-infected children. None the less, the use of combined antiretroviral drugs led to adequate viral suppression.