To analyze the influencing factors of ineffective treatment with lamotrigine combined with sodium valproate in patients with post-stroke epilepsy. Methods: The clinical data were collected from 269 patients with post-stroke epilepsy in the Zhongnan Hospital of Wuhan University between March 2020 and April 2023. According to curative effect of lamotrigine combined with sodium valproate, patients were divided into effective group (n=215) and ineffective group (n=54). Among them, 219 cases were randomly selected as the training set and the other 50 cases were selected as the verification set. The patients in effective group and ineffective group were matched by propensity score matching method. In data set of balanced covariates, influencing factors of the curative effect were analyzed by Logistic regression model. The prediction model for the curative effect was constructed, and its predictive efficiency was evaluated by receiver operating characteristic (ROC) curves. Results: In the 219 patients with post-stroke epilepsy after 1 ∶ 1 propensity score matching method, there were 42 pairs of patients with successful matching, and the data such as age, gender, underlying diseases, smoking history and drinking history were basically balanced. The proportions of NIHSS score >14 points and monthly onset frequency ≥4 times, levels of homocysteine and uric acid in effective group were lower than those in ineffective group (P<0.05). The results of Logistic regression analysis showed that NIHSS score >14 points, monthly onset frequency ≥4 times and high level of homocysteine were independent influencing factors of ineffective treatment (P<0.05). The prediction model was as follow: P=e X /(1+e X ), X=－8.423+0.532×NIHSS score +0.575× monthly onset frequency +0.436× homocysteine. Hsomer-Lemeshow goodness-fit test showed that there was no significant difference between predicted value of the model and the actual observed value (P>0.05). Internal verification of the prediction model showed that AUC, sensitivity and specificity were 0.818, 78.57% and 73.81%, respectively (P<0.05). External verification of the prediction model with validation set data showed that AUC, sensitivity and specificity were 0.815, 62.50% and 88.10%, respectively. Conclusion: NIHSS score >14 points, monthly onset frequency ≥4 times and high level of homocysteine are independent influencing factors of ineffective treatment with lamotrigine combined with sodium valproate. The prediction model has good predictive efficiency for the curative effect.