In the follow-up of patients treated for high grade glioma, differentiation between progressive disease (PD) and treatment-induced necrosis (TIN) is challenging. The purpose of this study is to evaluate the diagnostic accuracy of FDG PET for the differentiation between TIN and PD after high grade glioma treatment. We retrospectively identified patients between January 2011 and July 2013 that met the following criteria: age >18; glioma grade 3 or 4; treatment with radiotherapy or chemoradiotherapy; new or progressive enhancement on post treatment MRI; FDG PET within 4 weeks of MRI. Absolute and relative (to contralateral white matter) values of SUVmax and SUVpeak were determined in new enhancing lesions on MRI. The outcome of PD or TIN was determined by neurosurgical biopsy/resection, follow-up MRI, or clinical deterioration. The association between FDG PET and outcome was analyzed with univariate logistic regression and ROC analysis for: all lesions, lesions >10, >15, and >20 mm. We included 30 patients (5 grade 3 and 25 grade 4), with 39 enhancing lesions on MRI. Twenty-nine lesions represented PD and 10 TIN. Absolute and relative values of SUVmax and SUVpeak showed no significant differences between PD and TIN. ROC analysis showed highest AUCs for relative SUVpeak in all lesion sizes. Relative SUVpeak for lesions >20 mm showed reasonable discriminative properties [AUC 0.69 (0.41–0.96)]. FDG PET has reasonable discriminative properties for differentiation of PD from TIN in high grade gliomas larger than 20 mm. Overall diagnostic performance is insufficient to guide clinical decision-making.