Background The clinical application of flash glucose monitoring system (FGM) is becoming increasingly widespread, which can be continuously monitored for 14 days and does not require fingertip blood correction during wearing, providing a basis for clinical treatment. Therefore, the accuracy of FGM is particularly important. Objective To investigate the effect of human body composition and serum biochemical indicators on the accuracy of FGM sensor. Methods Patients with type 2 diabetes mellitus (T2DM) hospitalized in the Department of Endocrinology of the Second Affiliated Hospital of Zhengzhou University in 2022 were selected as the study objects, and their general data was collected through the electronic medical record system. The blood glucose was detected by FGM and compared with intravenous blood glucose, the physical analysis data of patients was collected. Fasting venous blood was collected to analyze hematological parameters. The clinical accuracy of FGM was evaluated by Clarke error grid analysis. The included patients were divided into the accurate group (MARD<10%, n=23) and inaccurate group (MARD>20%, n=34) according to the matched mean absolute relative difference of blood glucose (MARD). Binary Logistic regression analysis was used to analyze the influencing factors of FGM accuracy. Results A total of 694 pairs of blood glucose data were collected. Clarke error grid analysis was performed on the blood glucose values of FGM scans using venous blood glucose as the reference value. The results showed that 82.9% fell in zone A, 16.9% fell in zone B, 99.8% fell in zone A+B, and 0.2% fell in zone D, with an average MARD of 12.7%. MARD and muscle mass in the inaccurate group were higher than those in the accurate group (P<0.05), while uric acid, body fat mass and fat percentage were lower than those in the accurate group (P<0.05). Uric acid, body fat mass and fat percentage of male patients in the inaccurate group were lower than those in the accurate group, while the MARD value and muscle mass of male patients were higher than those in the accurate group (P<0.05). MARD value of female patients in the inaccurate group was higher than that in the accurate group (P<0.05). Binary Logistic analysis showed that muscle mass and blood uric acid concentration were influencing factors of the accuracy of FGM (P<0.05) . Conclusion The overall accuracy of FGM sensor meets the international standards. The accuracy of FGM sensor is related to uric acid level and muscle mass, but it was not affected by electrolyte ions in blood and other biochemical indicators, and interfered by human moisture, fat mass, inorganic salt content, fat thickness of the sensor wearing site and other factors.