Background: Angiogenesis is crucial for glioblastoma growth, and anti-vascular endothelial growth factor agents are widely used in recurrent glioblastoma patients. The number of circulating endothelial cells (CECs) is a surrogate marker for endothelial damage. We assessed their kinetics and explored their prognostic value in patients with recurrent glioblastoma. Methods: In this side study of the BELOB trial, 141 patients with recurrent glioblastoma were randomised to receive single-agent bevacizumab or lomustine, or bevacizumab plus lomustine. Before treatment, after 4 weeks and after 6 weeks of treatment, CECs were enumerated. Results: The number of CECs increased during treatment with bevacizumab plus lomustine, but not during treatment in the single-agent arms. In patients treated with lomustine single agent, higher absolute CEC numbers after 4 weeks (log(10)CEC hazard ratio (HR) 0.41, 95% CI 0.18-0.91) and 6 weeks (log(10)CEC HR 0.16, 95% CI 0.05-0.56) of treatment were associated with improved overall survival (OS). Absolute CEC numbers in patients receiving bevacizumab plus lomustine or bevacizumab single agent were not associated with OS. Conclusion: CEC numbers increased during treatment with bevacizumab plus lomustine but not during treatment with either agent alone, suggesting that this combination induced the greatest vascular damage. Although the absolute number of CECs was not associated with OS in patients treated with bevacizumab either alone or in combination, they could serve as a marker in glioblastoma patients receiving lomustine single agent.