Predictive markers and prognostic models are useful for the individualization of cancer treatment. In this study, we sought to identify clinical and molecular factors to predict overall survival in recurrent glioma patients receiving bevacizumab-containing regimens. A cohort of 102 patients was retrospectively collected from June 2011 to January 2022 at our institution. A nomogram was generated by Cox regression and feature selection algorithms based on 19 clinicopathological and 60 molecular variables. The model's performance was internally evaluated by bootstrapping in terms of discrimination and calibration. The median overall survival from the initiation of bevacizumab administration to death or last follow-up was 11.6 months (95% CI: 9.2–13.8 months) for all 102 patients, 10.2 months (95% CI: 6.4–13.3 months) for 66 patients with grade 4 tumors, and 13.8 months (lower limit of 95% CI: 11.5 months) for 36 patients with tumors of grade lower or not available. In the final model, a lower WHO 2021 grade (Grade lower or not available vs. Grade 4, HR: 0.398, 95% CI: 0.223–0.708, p = 0.00172), having received adjuvant radiochemotherapy (Yes vs. No, HR: 0.488, 95% CI: 0.268–0.888, p = 0.0189), and wildtype EGFR (Wildtype vs. Altered, HR: 0.193, 95% CI: 0.0506–0.733, p = 0.0157; Not available vs. Altered, HR: 0.386, 95% CI: 0.184–0.810, p = 0.0118) were significantly associated with longer overall survival in multivariate Cox regression. The overall concordance index was 0.652 (95% CI: 0.566–0.714), and the areas under the time-dependent curves for 6-, 12-, and 18-month overall survival were 0.677 (95% CI: 0.516–0.816), 0.654 (95% CI: 0.470–0.823), and 0.675 (95% CI: 0.491–0.860), respectively. A prognostic model for overall survival in recurrent glioma patients treated with bevacizumab-based therapy was established and internally validated. It could serve as a reference tool for clinicians to assess the extent the patients may benefit from bevacizumab and stratify their treatment response.