Abstract Purpose In the randomized phase III trial CeTeG/NOA-09, temozolomide (TMZ)/lomustine (CCNU) combination therapy was superior to TMZ in newly diagnosed MGMT methylated glioblastoma, albeit reporting more frequent hematotoxicity. Here, we analyze high grade hematotoxicity and its prognostic relevance in the trial population. Methods Descriptive and comparative analysis of hematotoxicity adverse events ≥ grade 3 (HAE) according to the Common Terminology of Clinical Adverse Events, version 4.0 was performed. The association of HAE with survival was assessed in a landmark analysis. Logistic regression analysis was performed to predict HAE during the concomitant phase of chemotherapy. Results HAE occurred in 36.4% and 28.6% of patients under CCNU/TMZ and TMZ treatment, respectively. The median onset of the first HAE was during concomitant chemotherapy (i.e. first CCNU/TMZ course or daily TMZ therapy), and 42.9% of patients with HAE receiving further courses experienced repeat HAE. Median HAE duration was similar between treatment arms (CCNU/TMZ 11.5; TMZ 13 days). Chemotherapy was more often discontinued due to HAE in CCNU/TMZ than in TMZ (19.7 vs. 6.3%, p = 0.036). The occurrence of HAE was not associated with survival differences (p = 0.76). Regression analysis confirmed older age (OR 1.08) and female sex (OR 2.47), but not treatment arm, as predictors of HAE. Conclusion Older age and female sex are associated with higher incidence of HAE. Although occurrence of HAE was not associated with shorter survival, reliable prediction of patients at risk might be beneficial to allow optimal management of therapy and allocation of supportive measures. Trial registration NCT01149109.