Clinical database construction has made great achievements in clinical research of thyroid tumors. Large database-based clinical studies lead the development and updating of guidelines, and promote the development of clinical practice in thyroid surgery. The thyroid cancer database construction starts relatively late in China. There is still a lack of high-quality evidence-based medicine data on Chinese population for thyroid cancer, as well as reference evidence for clinical guidelines. In recent years, the construction of thyroid tumor clinical database has increasingly developed depending on the evolution of internet technology and a large number of patients in China. However, due to the lack of quality control standards for database construction, and with the improvement of data availability, the risk of authenticity and integrity reduction is greatly increased during the process of various clinical information digitalizing. Based on experience and problems encountered during construction of thyroid tumor clinical database in the authors’ center, the authors present some preliminary ideas and suggestion on graded quality management of thyroid tumor clinical database. The process of database construction and application can be summarized as extracting the original information to convert the data for the final scientific research paper. The information characteristics of database were summarized as a step by step process of data acquisition on a scale of 0 to 3. The quality control of the data depends on the authenticity and availability of the data at the superior adjacent level. With the progressively transcriptions of information data, the data availability gradually increased to the level for scientific research application, while the authenticity gradually decreased. Database quality control is to ensure the authenticity of data as much as possible in the process of improving availability. Improving level-1 data quality is the foundation and key to improve the overall quality of the database, because that is at a higher risk of data loss during database construction. In order to improve the overall quality of the database, this paper shares some thoughts on data quality management from different aspects of medical record, including laboratory examination (level-1A), pathological report (level-1A), ultrasonic report (level-1B), and written record. In addition, we also propose structured level-1 medical record data recommendations. Over the years, clinical research of artificial intelligence diagnosis and disease prediction models based on big data has been carried out in full swing. However, the lack of satisfactory original database is difficult to improve later; the construction of a large database may lead to the embarrassing situation of “getting half the result with twice the effort”. This management problem may be ubiquitous in every specialized field and every national database construction. Therefore, the purpose of this paper is to call for the domestic colleagues pay attention to the quality control of the database and integrate the experiences of database construction from different centers, so as to brainstorm ideas and jointly develop a more reasonable hierarchical management of thyroid tumor clinical database in China.