This phase II study evaluated the efficacy and safety of AMG 102 (rilotumumab), a fully human monoclonal antibody against hepatocyte growth factor/scatter factor (HGF/SF), in patients with recurrent glioblastoma (GBM). Patients with histologically confirmed, measurable recurrent GBM or gliosarcoma (World Health Organization grade 4) and ≤3 relapses or prior systemic therapies received AMG 102 (10 or 20 mg/kg) by infusion every 2 weeks. The primary endpoint was best confirmed objective response rate (central assessment) per Macdonald criteria. Of the 61 patients who enrolled, 60 received AMG 102. Twenty-nine patients (48%) had previously received bevacizumab. There were no objective responses per central assessment, but 1 patient had an objective response per investigator assessment. Median overall survival (95% CI) in the 10- and 20-mg/kg cohorts was 6.5 months (4.1-9.8) and 5.4 months (3.4-11.4), respectively, and progression-free survival (PFS) per central assessment was 4.1 weeks (4.0-4.1) and 4.3 weeks (4.1-8.1), respectively. PFS was similar among patients who had previously received bevacizumab compared with bevacizumab-naive patients. The most common adverse events were fatigue (38%), headache (33%), and peripheral edema (23%). AMG 102 serum concentrations increased approximately dose-proportionally with 2-fold accumulation at steady state. Plasma total HGF/SF and soluble c-Met concentrations increased 12.05- and 1.12-fold, respectively, from baseline during AMG 102 treatment. AMG 102 monotherapy at doses up to 20 mg/kg was not associated with significant antitumor activity in heavily pretreated patients with recurrent GBM.