Objective To explore the predictive value of modified version Essen stroke risk score (ESRS) for recurrent cerebral infarction in 1 year. Methods Factors ofcerebrovascular stenosisand≥ 2 lesions in diffusion weighted imaging (DWI) which reflected large vascular stenosis and unstable plaques to the ESRS were added, and the 2 factors of other cardiovascular diseases and“peripheral vascular disease”which were difficult to operate clinically and had a lower incidence compared with the European and American people were got rid of. Thus, the total score of the modified version ESRS was still 9 scores. The risk of recurrent cerebral infarction in 1 year in 263 patients with cerebral infarction were evaluated by ESRS and modified version ESRS, and the predictive value was evaluated by area under receiver-operating characteristics (ROC) curve. Results The incidence of recurrent cerebral infarction in 1 year in 263 patients with cerebral infarction was 15.59% (41/263). There were statistical differences in the incidences of recurrent cerebral infarction in 1 year for different risk stratification patients by 2 methods (P<0.01). Multiple factor Logistic regression analysis results showed that thecerebrovascular stenosisand≥2 lesions in DWIwere the independent risk factors in predicting recurrent cerebral infarction in 1 year (OR=12.48 and 18.72, 95%CI 5.083-30.641 and 7.718-43.242, P=0.000 and 0.000). The area under ROC of predicting recurrent cerebral infarction in 1 year by ESRS and modified version ESRS was 0.68 (95% CI 0.58- 0.79) and 0.70 (95% CI 0.60-0.79), and there was statistical difference (P<0.01). Conclusions The predictive value of modified version ESRS in recurrent cerebral infarction in 1 year is higher than that of ESRS, and it is an effective method.