Objective To explore the influencing factors of cognitive dysfunction in patients with primary cerebral hemorrhage and establish a predictive model and validate it. Methods Using convenience sampling method， 195 patients with primary cerebral hemorrhage who were admitted to the Department of Neurosurgery， the First Affiliated Hospital of Nanjing Medical University from June 2019 to June 2022 were retrospective analyzed. All patients were followed up for over 6 months after discharge. All patients were assessed by Montreal Cognitive Assessment （MoCA） and Mini-Mental State Examination （MMSE） for cognitive function. The patients were divided into the cognitive dysfunction group （n=120） and the non-cognitive impairment group （n=75）. Binomial Logistic regression analysis was used to determine the influencing factors for cognitive dysfunction in patients with primary cerebral hemorrhage. According to the results of the analysis， a nomogram prediction model was constructed. The consistency coefficient （C-index） and calibration curve were used to evaluate the prediction efficiency and compliance of the nomogram model. Results The differences in age， education， National Institutes of Health Stroke Scale （NIHSS） score at the time of discharge， comorbid epilepsy，comorbid debridement， comorbid insular sign or high number of satellite foci， status of intensive caregiver， and distribution of bleeding sites were statistically significant when compared between the two groups （P ＜ 0.05）. Binomial Logistic regression analysis showed that age ＞ 60 years old （OR=4.689， 95%CI=1.420 to 15.470， P=0.011）， history of epilepsy （OR=3.007， 95%CI=1.270 to 7.118， P=0.012）， NHISS score ＞ 15 points at discharge （OR=2.699， 95%CI=1.257 to 5.797， P=0.011）， and the hemorrhage site were risk factors of cognitive dysfunction after primary cerebral hemorrhage. Primary school education level or below （OR=0.382， 95%CI=0.159 to 0.913， P=0.030）， the site of hemorrhage being the parieto-occipital lobe （OR=0.105， 95%CI=0.019 to 0.579， P=0.010） was a protective factor for patients with primary cerebral hemorrhage with concomitant cognitive deficits. The results of the model validation showed that，the C-index of the nomogram model was 0.785， the calibration curve is relatively consistent with the trend of the ideal curve， showed that the prediction model had good accuracy and consistency. Conclusions Patients with primary cerebral hemorrhage who are older， with lower educational level， higher discharged NHISS score， epilepsy history and hemorrhage in brain stem， basal ganglia thalamus region and frontotemporal region are prone to post-stroke cognitive dysfunction. The nomogram model constructed based on the above variables has high prediction efficiency.