Background: Both lomustine (LOM) and bevacizumab (BEV) are approved agents for recurrent or progressive glioblastoma. We aimed at evaluating the optimal treatment sequence (single agent vs sequential vscombination) of these agents in a randomized phase II design. Methods: Progressive patients after initial radiotherapy and temozolomide were randomized 2:2:2:1 between Arm 1: LOM 90 mg/m² (max. 160 mg) every 6 weeks plus BEV (10 mg/kg) every 2 weeks until progression, subsequent salvage treatment at the investigators best choice (MD choice); Arm 2: LOM 110 mg/m² (max. 200 mg) every 6 weeks, at progression switch to BEV every two weeks; Arm 3: BEV every 2 weeks, at progression add LOM 90 mg/m² (max. 160 mg) every 6 weeks and continue BEV; and Arm 4: LOM single agent 110 mg/m² (max. 200 mg) every 6 weeks, at progression MD choice (arm 4). Eligibility criteria included performance status 0-2, adequate hematological, liver and renal function, and an interval since the end of radiotherapy of ≥ 3 months to rule out pseudoprogression. The primary endpoint was overall survival at 12 months (OS12), progression free survival (PFS) was measured from the start of study treatment until 1^st PD. Response was locally assessed using RANO criteria. Results: Between Oct 2011 and July 2013 274 patients were randomized. Baseline characteristics were well balanced, and all treatment generally well tolerated. In arm 4, 43% received subsequent salvage BEV. Time to first progression was longer in initially BEV treated patients, but survival was similar in all arms. Conclusions: Initially BEV treated patients had longer PFS, but OS 12 or median OS were similar in the various schedules of BEV and LOM. Following the emergence of the data of the BELOB study, prior to any analysis arms 1 and 4 of this phase II trial were continued as a phase III trial. Clinical trial information: NCT01290939