Objective: To investigate the correlation between legumain(LGMN) and early neurological deterioration（END） in patients with acute large artery atherosclerotic（LAA） stroke. Methods: Patients with acute LAA stroke admitted at the Department of Neurology, Shidong Hospital of Yangpu District in Shanghai from January 2016 to December 2018 were enrolled. Serum levels of LGMN and other related laboratory indices were tested in the patients. END was defined as an increase of at least 2 points in repeated National Institutes of Health Stroke Scale (NIHSS) score within the first 72 h of hospitalization. According to whether the patients had END or not, they were divided into END group and non-END group. Multivariable logistic regression model was used to detect the independent risk factors for END. Receiver operator characteristic (ROC) curve was used to analyze the value of LGMN in predicting the occurrence of END. Results: A total of 334 patients were included, and 78 patients (23.4%) had END (END group), while the other 256 patients were of the non-END group. Univariate analysis showed that the incidence of accompanied disease (hypertension, diabetes, hyperlipidemia), vulnerable plaque formation, systolic blood pressure on admission, level of LGMN, fasting blood glucose, glycated hemoglobin, hypersensitive C-reactive protein, NIHSS score on admission in END group were statistically higher than those in non-END group (all P<0.05). The serum LGMN level was significantly lower in moderate (7.05 μg/L) and occlusion (6.80 μg/L) groups than in severe stenosis group (8.66 μg/L) (P<0.05). Multivariate logistic regression analysis showed that increased level of LGMN (OR=1.309,95% CI 1.188-1.441,P<0.001) was the independent risk factors for the occurring of END; other risk factors included vulnerable plaque formation, high systolic blood pressure on admission, high glycated hemoglobin and NIHSS score on admission. ROC curve analysis showed that the area under the curve of LGMN for predicting END was 0.715 (95% CI: 0.663-0.763, P<0.001), with an optimal cut-off value of 6.73 μg/L. Conclusions: Increased level of LGMN is an independent risk factor for END in patients with LAA stroke, which may be used as a biomarker for predicting END in patients with the disease.