Background: Class sparing antiretroviral therapy with only protease inhibitors (PI) had been associated with central nervous system (CNS)-replication of HIV-1, probably due to impaired CNS penetration-effectiveness, but clinical confirmation is scarce. Objective was to evaluate risk factors for CNS replication of HIV-1 in patients on only boosted PI, versus nucleoside reverse transcriptase inhibitor (NRTI)-containing triple therapy. Methodology: Retrospective case-control study (1:5-ratio), with patients from a large German HIV-treatment unit, data were collected from 2005- 2013. All patients with available data on antiretroviral therapy, clinical and demographic characteristics, and paired results from simultaneous HIV-1 load-assessments in cerebrospinal fluid (CSF) and blood plasma, were eligible. CSF was collected for various reasons, patients were both neurologically symptomatic, or asymptomatic. A virus-load (VL)-ratio for CSF/blood-quotient was built, and patients were evaluated according to treatment with either only boosted PI (cases), or NRTI-containing triple therapy (controls). Univariate group statistics and Pearson’s regression analysis for time-dependent results for PI-patients were generated. Results: 155 patients were evaluated: 24 cases and 131 controls. All cases used boosted double PI-regimens, consisting of saquinavir (21), lopinavir (20), atazanavir (4), fosamprenavir (2) and indinavir (1), respectively. At time of CSF collection, the mean age was 47 years (y), 22% were female, time from first HIV-diagnosis was 10.2y (cases 15.7 vs controls 9.2, p<0.001), the mean number of regimens was 7.39 for cases and 4.24 for controls (p=0.002), and median CD4 count were 186/mm3, or 162/mm3 (n.s.). The proportion of patients with undetectable HIV-1 in CSF was 25% for cases and 49.6% for controls (p=0.026) whereas in blood 46% vs. 41% (n.s.). Median CSF-VL was 600 cop/mL for cases, and 50 cop/mL in controls (p=0.027); median VL from blood was 115cop/mL, or 173cop/mL (n.s.). Mean VL-ratio was 342.91 for cases and 54.48 for controls (p=0.002). Pearson’s regression analysis showed an increasing trend for VL-ratio over time for cases (see figure). Conclusions: In this unselected cohort, HIV-1 replication in CSF was higher for patients on only PI, than for those on NRTI-containing triple therapy, supporting a worse CNS-penetration effectiveness for used PI. Patients’ over time rise in CNS-viral replication suggests that this effect was time-dependent.