ABSTRACT Background The potential benefits of 18F‐fluoro‐2‐deoxy‐D‐glucose‐positron emission tomography/CT (FDG‐PET/CT) imaging for radiotherapy (RT) treatment planning of head and neck squamous cell carcinoma (HNSCC) are increasingly being recognized. It has been suggested that intratumoral subvolumes with high FDG avidity (“hotspots”) are potential targets for selected dose escalation. The purposes of this study were to demonstrate that pre‐RT FDG‐PET/CT can identify intratumoral sites at increased risk of local relapse after RT and to determine an optimal threshold to delineate smaller RT target volumes that would facilitate RT dose escalation without impaired tolerance. Methods Seventy‐two consecutive patients with locally advanced HNSCC treated by RT ± chemotherapy were included in this study. All patients underwent FDG‐PET/CT at initial staging (PETA) and during systematic follow‐up (PETR). FDG‐PET/CT was coregistered on the initial CT scan with a rigid method. Various subvolumes (AX; × = 30%, 40%, 50%, 60%, 70%, 80%, and 90% standardized uptake value maximum SUVmax thresholds) within the primary tumor and in the subsequent local relapse (RX; × = 40% and 70% SUVmax thresholds) were compared together (Dice, Jaccard, overlap fraction, common volume/baseline volume, and common volume/recurrent volume). Results Nineteen patients (26%) had local relapses. Using a 40% SUVmax threshold, the initial metabolic tumor volume was significantly higher in patients with local relapses than in controlled patients (10.4 ± 8.6 vs 5.1 ± 4.9 cc; p = .002) as well as total lesion glycolysis (117.9 ± 88.6 vs 60.6 ± 80.4; p = .013). For both methods, the overlap index among A30, A40, and A50 subvolumes on PETA and the whole metabolic volume of recurrence R40 and R70 on PETR showed a moderate agreement (0.52 to 0.43). Conclusion Our study does not find high overlap index values between the initial tumor and recurrence subvolumes, probably because of a suboptimal coregistration. Our results also confirm that metabolic tumor volume and total lesion glycolysis are independently correlated with recurrence‐free survival in patients with HNSCC. Further larger prospective studies with FDG‐PET/CT performed in the same RT position and with a validated elastic registration method are needed. © 2017 Wiley Periodicals, Inc. Head Neck 39: 1155–1165, 2017