Objective To investigate the correlations of plasma brain natriuretic peptide （BNP） levels and the subtypes and the severity of acute ischemic stroke. Methods Consecutive patients with acute ischemic stroke were enrolled in the study. They were divided into large-artery atherosclerosis （LAA）, small- artery occlusion （SAO） and cardioembolism （CE） according to the Trial of Org 10172 in Acute Stroke Treatment （TOAST） classification. Electrochemiluminescence immtmoassay was used to measure plasmaBNP. OsiriX software was used to calculate the cerebral infarction volume of MRI. Results A total of 159 patients with acute ischemic stroke were enrolled. Compared to the CE group, there were significant differences in the proportions of hyperlipidemia, atrial fibrillation, BNP 〉 100 ng/L, and in the levels of low- density lipoprotein cholesterol （LDL-C） and BNP, as well as in the National Institutes of Health Stroke Scale （NIHSS） score and the infarct volume in the groups of LAA and SAO （all P 〈 0. 05）. Multivariate logistic regression analysis showed that atrial fibrillation （odds ratio [ OR] 148. 638, 95% confidence interval [ CI] 17. 532-1 260.182;P〈0.001） and BNP 〉100ng/L （OR 3. 490, 95%CI 1.033-11.789; P=0.044） were the independent risk factors for CE ischemic stroke. Receiver operating characteristic curve analysis showed that the optimal cutoff value for predicting plasma BNP level of the CE ischemic stroke was 101.4 ng/L, and its sensitivity, specificity, positive predictive value, and negative predictive values were 80. 56%, 76. 42%, 50.00%, and 93.07%, respectively. The optimal cutoff value was used as a boundary, the patients were divided into a BNP ≤ 101.4 ng/L group and a 〉 101.4 ng/L group. The proportions of males （P =0. 031 ）, hyperlipidemia （P= 0. 002） and smoking （P= 0. 026）, as well as the levels of total cholesterol （P= 0. 020）, triglyceride （P=0. 024） and LDL-C （P= 0. 013） were significantly lower than those in the BNP≤ 101.4 ng/L group, while the proportions of patients with ischemic heart disease （P 〈 0.001） and atrial fibrillation （P〈 0.001）, as well as the NIHSS score and the infarct volume were significantly higher and larger than those in the BNP ≤101.4 ng/L group. Spearman rank correlation analysis showed that the plasma BNP levels were significantly negatively correlated with the triglycerides （r= -0.224, P= 0.004） and LDL-C （r= -0. 170, P= 0. 032） levels, and were significantly positive correlated with the NIHSS scores （P= 0. 167, P=0. 044） and the infarct volume （P=0. 281, P 〈 0. 001）. Conclusions Plasma BNP level can be used as a biological marker for early differentiating CE from non-CE ischemic stroke, as well as identifying the severity of stroke.