OBJECTIVE： To construct prediction model for initial dose of levothyroxine （L-T4） in differentiated thyroid cancer （DTC） patients after surgery. METHODS： A total of 100 DTC patients underwent surgery were selected from thyroid and breast surgery department in Nanjing Drum Tower Hospital. General information of patients such as gender， age， height， body mass， body mass index （BMI） and regular follow-up information after discharge were collected. Related data of thyroid function and adjusted dose of L-T4 were recorded. Single factor variance analysis and t-test were used to analyze the predictors that had significant correlation with the initial dose of L-T4. The prediction model of L-T4 initial dose was established by linear regression analysis， and was verified by prospective experiments. RESULTS： The initial dose of L-T4 in DTC post-surgery patients were significantly correlated with age （P＝0.01，F＝3.993）， body weight （P＜0.001，F＝6.910） and BMI （P＜0.001，F＝7.698）. Linear regression analysis showed that prediction model of initial dose of L-T4 was L-T4（μg/kg）＝2.971-0.033×BMI-0.005×age. DTC post-operative patients were given L-T4 empirically， and only 16％ （16/100） of the patients met the criteriaat the first follow-up  of thyroicl function. In the validation test， L-T4 was given at the initial dose calculated by the prediction model， and 63.7％      （44/69） of the patients reached the standard at the first follow-up. CONCLUSIONS： The established prediction model of L-T4 initial dose after DTC surgery has a certain practicality.