The adequate risk stratification in thyroid carcinoma is crucial to avoid on one hand the overtreatment of low-risk and on the other hand the undertreatment of high-risk patients. The question how to properly assess the risk of relapse has been discussed during recent years and resulted in a substantial change in our approach to risk stratification in differentiated thyroid cancer, proposed by the newest ATA guidelines. First initial risk stratification, based on histopathological data is carried out just after primary surgery. It should be emphasized, that a high quality of histopathological report is crucial for proper risk stratification. Next, during the follow-up, patients are restratified considering their response to treatment applied and classified to one of the following categories: excellent response, biochemical incomplete response, structural incomplete or indeterminate response. This new approach is called dynamic risk stratification as, in contrary to the previous rigid evaluation performed at diagnosis, reflects a real-time prognosis and thereby substantially influences and personalizes disease management. In this review, we raise some unresolved questions, among them the lack of prospective studies, fulfilling evidence-based criteria, necessary to validate this model of risk stratification. We also provided some data concerning the use of dynamic risk stratification in medullary thyroid cancer, not yet reflected in ATA guidelines. In conclusion, dynamic risk stratification allows for better prediction of the risk of recurrence in thyroid carcinoma, what has been demonstrated in numerous retrospective analyses. However, the validation of this approach in prospective studies seems to be our task for near future.