Background: Glioblastoma (GBM) is one of the worst prognostic cancers and there is no effective treatment after failure of bevacizumab. Eribulin is a microtubule inhibitor used for the treatment of patients with metastatic breast cancer and liposarcoma. We previously reported that eribulin strongly inhibits the RNA-dependent RNA polymerase (RdRP) activity of TERT protein in cancer cells, and has a strong anti-tumor effect against GBM cells with TERT promoter mutation. In this study we aim to investigate the efficacy and safety of eribulin in patients with bevacizumab-resistant recurrent GBM. Methods: This is an open-label, multicenter, single-arm phase II trial. Eligible patients aged 20-75 years with bevacizumab-resistant recurrent GBM were enrolled from 2018-2020. Patients received eribulin 1.4 mg/m² on days 1 and 8 of 21-day cycle until disease progression or intolerable toxicity was observed. The primary endpoint was one-year overall survival rate (1yOS%). The 35 patients are needed to achieve an 80% power at a one-sided alpha of 10%, under threshold 1yOS% of 10% and expected 1yOS% of 25%. Results: Thirty-seven patients aged 26-73 (median: 54) years were treated. Twenty-six of 37 (70.3%) patients were diagnosed as IDH-wildtype GBM, 4 (10.8%) were with IDH-mutant GBM and 7 (18.9%) were GBM, NOS. Thirty-four (91.9%) patients had a Karnofsky performance status of 70 or 80 at the registration. Thirty-one (83.8%) patients received additional treatments, including 28 (75.7%) bevacizumab, 11 (29.7%) re-irradiation and 3 (8.1%) resection after failure of eribulin. Among 37 subjects, 32 surgical specimens were analyzed for TERT promoter mutation and 15 for RdRP activity. 1yOS% was 29.7% [80% CI: 20.5 to 39.5 (p < 0.0001), 95% CI: 16.1 to 44.6]. Median OS was 9.0 months [95% CI: 6.2 to 11.0] and median progression-free survival was 1.5 months [95% CI: 1.4 to 1.7]. Neither TERT nor RdRP statuses was associated with prolonged OS. Among all the target lesions evaluated, two lesions decreased more than 50% in size and the patients survived more than one year, however no obvious PR was confirmed at the final evaluation. The disease control rate was 25.7% [95% CI: 12.5 to 43.3]. Common ≥ grade 2 AEs were neutropenia (70.3%), leukopenia (56.8%), lymphopenia (27.0%), elevation of γ-GTP (13.5%), elevation of ALT (10.8%), elevation of AST (8.1%), alopecia (8.1%). Treatment-related grade 3 or 4 AEs occurred in 59.5% of subjects. There were no AEs leading to death. Conclusions: Eribulin was safely applied for the patients with recurrent GBM. This phase II study met its primary endpoint of 1yOS%, although no obvious response was observed. Further investigation to reveal the biomarkers related to longer survival is underway. Clinical trial information: UMIN000030359.