Limiting dietary carbohydrates inhibits glioma growth in preclinical models. Therefore, the ERGO trial (NCT00575146) examined feasibility of a ketogenic diet in 20 patients with recurrent glioblastoma. Patients were put on a low-carbohydrate, ketogenic diet containing plant oils. Feasibility was the primary endpoint, secondary endpoints included the percentage of patients reaching urinary ketosis, progression-free survival (PFS) and overall survival. The effects of a ketogenic diet alone or in combination with bevacizumab was also explored in an orthotopic U87MG glioblastoma model in nude mice. Three patients (15%) discontinued the diet for poor tolerability. No serious adverse events attributed to the diet were observed. Urine ketosis was achieved at least once in 12 of 13 (92%) evaluable patients. One patient achieved a minor response and two patients had stable disease after 6 weeks. Median PFS of all patients was 5 (range, 3-13) weeks, median survival from enrollment was 32 weeks. The trial allowed to continue the diet beyond progression. Six of 7 (86%) patients treated with bevacizumab and diet experienced an objective response, and median PFS on bevacizumab was 20.1 (range, 12-124) weeks, for a PFS at 6 months of 43%. In the mouse glioma model, ketogenic diet alone had no effect on median survival, but increased that of bevacizumab-treated mice from 52 to 58 days (p<0.05). In conclusion, a ketogenic diet is feasible and safe but probably has no significant clinical activity when used as single agent in recurrent glioma. Further clinical trials are necessary to clarify whether calorie restriction or the combination with other therapeutic modalities, such as radiotherapy or anti-angiogenic treatments, could enhance the efficacy of the ketogenic diet.