Abstract Background Patients with glioblastoma (GBM) have a dramatically poor prognosis. The recent REGOMA trial suggested an overall survival (OS) benefit of regorafenib in recurrent GBM patients. Considering the extreme genetic heterogeneity of GBMs, we aimed to identify molecular biomarkers predictive of differential response to the drug. Methods Total RNA was extracted from tumor samples of patients enrolled in the REGOMA trial. Genome-wide transcriptome and micro (mi)RNA profiles were associated with patients’ OS and progression-free survival. Results In the first step, a set of 11 gene transcripts (HIF1A, CTSK, SLC2A1, KLHL12, CDKN1A, CA12, WDR1, CD53, CBR4, NIFK-AS1, RAB30-DT) and 10 miRNAs (miR-93-5p, miR-203a-3p, miR-17-5p, let-7c-3p, miR-101-3p, miR-3607-3p, miR-6516-3p, miR-301a-3p, miR-23b-3p, miR-222-3p) was filtered by comparing survival between regorafenib and lomustine arms. In the second step, a mini-signature of 2 gene transcripts (HIF1A, CDKN1A) and 3 miRNAs (miR-3607-3p, miR-301a-3p, miR-93-5p) identified a subgroup of patients showing prolonged survival after regorafenib administration (median OS range, 10.6–20.8 mo). Conclusions The study provides evidence that a signature based on the expression of 5 biomarkers could help identify a subgroup of GBM patients exhibiting a striking survival advantage when treated with regorafenib. Although the presented results must be confirmed in larger replication cohorts, the study highlights potential biomarker options to help guide the clinical decision among regorafenib and other treatments in patients with relapsing GBM.