OBJECTIVES: The objectives were (i) to confirm that diagnostic FDG-PET images could be obtained during thoracic radiotherapy, (ii) to verify that significant changes in FDG uptake or volume could be measured early enough to adapt the radiotherapy plan and (iii) to determine an optimal time window during the radiotherapy course to acquire a single FDG-PET examination that would be representative of tumour response. METHODS: Ten non-small cell lung carcinoma (NSCLC) patients with significant PET/CT-FDG tumour radioactivity uptake (versus the background level), candidates for curative radiotherapy (RT, n=4; 60-70 Gy, 2 Gray per fraction, 5 fractions per week) or RT plus chemotherapy (CT-RT, n=6), were prospectively evaluated. Using a Siemens Biograph, 5 or 6 PET/CT scans (PET(n), n=0-5) were performed for each patient. Each acquisition included a 15-min thoracic PET with respiratory gating (RG) 60±5 min post-injection of the FDG (3.5 MBq/kg), followed by a standard, 5-min non-gated (STD) thoracic PET. PET(0) was performed before the first RT fraction. During RT, PET(1-5) were performed every 7 fractions, i.e., at 14 Gy total dose increment. FDG uptake was measured as the variation of SUV(max,PETn) versus SUV(max,PET0). Each lesions' volume was measured by (i) visual delineation by an experienced nuclear physician, (ii) 40% SUV(max) fixed threshold and (iii) a semi-automatic adaptive threshold method. RESULTS: A total of 53 FDG-PET scans were acquired. Seventeen lesions (6 tumours and 11 nodes) were visible on PET(0) in the 10 patients. The lesions were located either in or near the mediastinum or in the apex, without significant respiratory displacements at visual inspection of the gated images. Healthy lung did not cause motion artefacts in the PET images. As measured on 89 lesions, both the absolute and relative SUV(max) values decreased as the RT dose increased. A 50% SUV(max) decrease was obtained around a total dose of 45 Gy. Out of the 89 lesions, 75 remained visually identifiable during the entire course of treatment. The 40% fixed threshold and adaptive threshold methods failed to delineate otherwise visible lesions in 16/33 (48%) and 3/33 (9%) lesions, respectively. The failure rate increased with increasing RT doses. Restricting the analysis to the manually-defined volumes in 89 visible lesions, the relative volumes decreased with increased dose. CONCLUSIONS: FDG-PET images can be analysed during thoracic RT, given either alone or with chemotherapy, without disturbing radiation-induced artefacts. An average 50% decrease in SUV(max) was observed around 40-45 Gy (i.e., during week 5 of RT). The three delineation methods yielded consistent volume measurements before RT and during the first week of RT, while manual delineation appeared to be more reliable later on during RT.