Abstract Background Angiogenesis is one of the most distinctive hallmarks of glioblastoma (GBM). Although bevacizumab did not show to improve overall survival in phase 3 trials, it was approved by FDA and is often prescribed as off-label therapy in the recurrent clinical setting. The aim of this study is to evaluate the difference in terms of survival and safety between the 5 mg/m2 and 10 mg/m2 bevacizumab schedule in recurrent GBM. Material and methods All pts treated at Veneto Institute of Oncology from May 2013 to March 2022 were retrospectively reviewed. Major inclusion criteria were: histologically confirmed diagnosis of IDHwt GBM/IDHmut grade 4 astrocytoma (according to the WHO 2021 classification), relapse after first or subsequent line of therapy, treatment with bevacizumab at 5 mg/m2 or/and 10 mg/m2 every 2 weeks until progression/death or unacceptable toxicity. Bevacizumab was administered as off-label therapy. The treatment schedule was at physician’s discretion. RANO criteria and CTCAE v5.0 were used for response and toxicity assessment. Results 81 pts were enrolled. From starting bevacizumab the median follow-up was 10.9ms [95% CI 9.8-14.0] and median age was 53ys (range 18-81). 33 (41%) pts received the 5 mg/m2 schedule. Among them, 2 (6%) were IDHmut grade 4, 8 (24%) had ≥65ys and ECOG-PS was 0-1 in 16 (48%) and ≥2 in 17 (51%), respectively. MGMT was methylated in 15 of 30 (50%) evaluable pts. Median number of prior lines of treatment was 2 (range 1-4) and 30% of pts received bevacizumab at first recurrence. 28 (84.9%) pts were evaluable for response: 7 (21%) and 5 (15%) showed PR and SD. 48 pts received the 10 mg/m2 schedule: 5 (10%) were IDHmut grade 4 astrocytoma; 29 (60%) had an ECOG-PS of 0 or 1 and 4 (8%) had ≥65ys, MGMT was methylated in 20 of 44 (45%) evaluable pts. 36 (75%) pts received bevacizumab beyond the second line of therapy. 46 (96%) pts were evaluable for response: 6 (12%) had PR, 19 (39%) SD. mOS from the start of bevacizumab was 7.3ms (95% CI 4.3-6.4), mPFS was 4.4ms [95% CI 3.7 - 6.4]. At univariate analysis, pts who received the 5 mg/m2 or the 10 mg/m2 schedule had a mOS of 5.4 and 7.7ms (p=0.08); mOS for pts with ECOG-PS < or ≥2 was 9.0 and 5.4ms (p=0.04) while mOS for pts with <2 or ≥2 lines of therapy was 4.7 and 7.7ms (p=0.056). Age and type of the tumor were not statistically significant. At multivariate analysis, MGMT methylated status was the only factor statistically associated with OS (HR=0.48, 95% CI, p=0.002) and PFS (HR=0.33, 95% CI, p=0.001), while a number of prior lines of therapy ≥2 (HR=2.07, 95% CI, p=0.02) was significantly associated only with PFS. Grade 3-4 most common adverse events were hypertension (18%) in pts treated with 5 mg/m2 and hypertension (16%) and proteinuria (2%) in pts treated with 10 mg/m2. Conclusions Bevacizumab treatment with a dosage of 5 mg/m2 and 10 mg/m2 seems to give comparable outcome in terms of survival in recurrent GBM pts. No difference was demonstrated for safety.