Background: Bevacizumab(BEV)is widely used in recurrent glioblastoma,alone or in combination with other agents.There is however no well-controlled trial to support the use for this indication.Methods: In a three-arm Dutch multicenter randomized phase Ⅱ study(NTR 1929)patients were assigned to either BEV 10 mg/kg iv every 2 weeks,BEV 10 mg/kg iv every 2 weeks and 110 mg/m2 lomustine every 6 weeks,or lomustine 110 mg/m2every 6 weeks.Eligible were patients with histologically proven glioblastoma,with a first recurrence after chemo-irradiation with temozolomide,having concluded radiotherapy more than 3 months ago,with adequate bone marrow,renal and hepatic function,and WHO performance status(PS)0-2.Primary endpoint was 9 months overall survival(OS); P0 was set at 35%and P1 at 55%.Progression was defined using RANO criteria.A safety review after the first 10 patients in the combination arm was preplanned.Results: Between December 2009 and November 2011,153 patients were enrolled of whom 148 were considered eligible.Median age was 57 years(range,24-77)and median WHO PS was 1.With respect to prognostic factors groups were well balanced.After review of the safety cohort the dosage lomustine in the combination arm was lowered to 90 mg/m2 because of hematological toxicity(predominantly thrombocytopenia without symptoms).At this lower lomustine dose level the combination treatment was in general well tolerated.Outcome: see Table.Conclusions: In this first well-controlled study on BEV in recurrent glioblastoma with a primary OS endpoint,combination treatment with bevacizumab and lomustine met the prespecified criterion for further investigation in clinical trials,whereas both drugs given as single agent failed to meet this criterion.