Objective: The aim of this study was to compare the TNM staging system and the American Thyroid Association (ATA) recurrence risk classification in predicting radioiodine refractory disease (RRD) in differentiated thyroid cancer (DTC) and to analyze the correlation of stimulated thyroglobulin (Tg) levels and rate of Tg elevation with the standardized uptake value on 18F-fludeoxyglucose (FDG) PET/CT scan. Methods: RRD was indicated through the retrospective analysis of consecutive 18F-FDG PET/CT scans in DTC with stimulated Tg >10 ng/ml and negative 131I NaI whole-body scans (WBS). Tg elevation velocity was compared to the likelihood of a positive scan. The ATA recurrence risk and TNM staging system were compared to see which of them better predicted the subsequent development of RRD. Results: Fifty-eight of 636 subjects developed RRD: 52 papillary and 6 follicular thyroid cancer. The median time between diagnosis and a negative WBS was 24 months (range 12-240). RRD developed in 11 low-risk, 32 intermediate-risk and 15 high-risk patients. A better response to therapy was seen in the low-risk versus the intermediate- and high-risk groups. 18F-FDG PET/CT scans had a diagnostic accuracy of 94.8%, sensitivity of 97.7%, specificity of 85.7%, positive predictive value of 95.6% and negative predictive value of 92%. There was no correlation between the Tg level or rate of rise and a positive scan. Overall, PET-CT upstaged 18 (31%) cases, leading to a change in management in 20 (35%) cases. Conclusion: The TNM and ATA staging systems show no significant difference in predicting the development of RRD. RRD is less likely in stage I, II and low-risk patients. There is no correlation between the level or rate of Tg rise and a positive 18F-FDG PET/CT scan.