Abstract Purpose External ventricular drain (EVD) is one of the most frequent procedures in neurosurgery and around 15 to 30% of these patients require a permanent cerebrospinal fluid (CSF) diversion. The optimal EVD weaning strategy is still unclear. Whether gradual weaning compared to rapid closure, reduces the rate of permanent CSF diversion remains controversial. The aim of this trial is to compare the rates of permanent CSF diversion between gradual weaning and rapid closure of an EVD. Methods This was a single-center, retrospective cohort study including patients between 2010 to 2020. Patients were divided into a weaning (WG) and non-weaning (NWG) group. The primary outcome was permanent CSF diversion rates, secondary outcomes included hospitalization time, EVD-related morbidity, and clinical outcome. Results Out of 412 patients, 123 (29.9%) patients were excluded due to early death or palliative treatment. We registered 178 (61.6%) patients in the WG and 111 (38.4%) in the NWG. Baseline characteristics were comparable between groups. The VPS rate was comparable in both groups (NWG 37.8%; WG 39.9%, p = 0.728). EVD related infection (13.5% vs 1.8%, p < 0.001), as well as non-EVD related infection rates (2.8% vs 0%, p < 0.001), were significantly higher in the WG. Hospitalization time was significantly shorter in the NWG (WG 24.93 ± 9.50 days; NWG 23.66 ± 14.51 days, p = 0.039). Conclusion Gradual EVD weaning does not seem to reduce the need for permanent CSF diversion, while infection rates and hospitalization time were significantly higher/longer. Therefore, direct closure should be considered in the clinical setting.