Background: It has previously been shown that bevacizumab, when added to chemotherapy, improved overall survival in several cancers. In glioblastoma multiforme (GBM), bevacizumab increased progression-free survival and it is widely used for tumor recurrence, though it has failed to improve overall survival (OS) in controlled trials. However, an effective biomarker for predicting the prognosis of bevacizumab treatment has yet to be identified. This study, therefore, aimed to retrospectively analyze the polymorphisms of p53 codon 72 and the clinical characteristics of GBM specimens from Taiwanese patients. Methods: The polymorphisms of p53 codon 72 in 99 patients with GBM treated at Taichung Veterans General Hospital in Taiwan from 2007 to 2017 were analyzed using direct DNA sequencing and PCR-RFLP analysis. Results: We found that among these GBM patients, the distribution of codon 72 polymorphisms was 28.3% for proline homozygotes (Pro/Pro), 38.4% for arginine homozygotes (Arg/Arg), and 33.3% for proline/arginine heterozygotes (Pro/Arg). Although the polymorphisms of p53 codon 72 were not directly associated with the overall survival of GBM, both the Arg/Arg and Arg/Pro genotypes were associated with significant benefits in terms of overall survival in patients treated with CCRT plus bevacizumab compared to patients treated with CCRT alone. Conclusions: This pilot study suggests that both the Arg/Arg and Arg/Pro genotypes of p53 codon 72 polymorphism may have value as independent prognostic or predictive parameters for bevacizumab treatment response and failure. Relatedly, the results of the study further demonstrate the utility of stratifying GBM patients according to bevacizumab sensitivity.