Purpose To evaluate the prognostic value of 18F-FDG-PET-CT performed prior to (prePET) and during the third week (iPET) of radiation therapy (RT) in nasopharyngeal carcinoma (NPC). Materials and methods Thirty-patients with newly diagnosed loco-regionally advanced NPC treated with radical RT underwent prePET and iPET. The median follow-up was 26 months (8–66.9). The maximum-standardised-uptake-value (SUVmax), metabolic-tumour-volume (MTV) and total-lesional-glycolysis (TLG) of the primary tumour (PT), index-node (IN) (lymph node with highest TLG), total-lymph-nodes (TN) and combined primary-tumour and nodal (PTN), and their % reductions in iPET were analysed, and results were correlated with 2-year Kaplan–Meier loco-recurrence-free-survival (LRFS), regional-failure-free-survival (RFFS), distant-metastatic-failure-free-survival (DMFFS), disease-free-survival (DFS), and overall-survival (OS). Optimal-cutoffs (OC) were derived from Receiver-Operating-Characteristic curves. Results For LRFS, the only predictor was reduction in PT MTV by >50%: 95.2% vs. 75.0%, p = 0.024. For other treatment outcomes, only nodal or PTN predicted outcomes. The IN SUVmax (pre-PET-OC = 10.45 g/mL and iPET-OC = 8.15) and TLG (prePET-OC = 90 g and iPET-OC = 33.4) were the best predictors of outcome: RFFS (iPET SUVmax/TLG): 100% vs. 50%, p < 0.001 and 100% vs. 44%, p = 0.032; DMFFS (prePET SUVmax/TLG); 100% vs. 51.9%, p = 0.004 and 100% vs. 47.6%, p = 0.002; DFS (prePET TLG and iPET SUVmax): 87.5% vs. 33%, p = 0.045 and 78.7% vs. 20%, p = 0.01; and OS (prePET TLG): 100% vs 66.3%, p = 0.036. Conclusions We have demonstrated IN of prePET and iPET to be a feasible and potentially useful novel imaging biomarker to predict for patients with NPC who have a high risk of regional or distant metastatic failure. Future work is required to validate our findings in a well-powered, prospective study with a standardised treatment protocol, and their potential use to guide individualised therapy for NPC.