Objective To analyze the significance of serum miR-145 and miR-210 expression levels in the diagnosis of carotid artery stenosis. Methods During the same period, 55 healthy individuals who received physical examination in the same hospital were recruited as controls and assigned to a non-stenosis group. Among the included patients, there were 45 cases of mild stenosis, 14 cases of moderate stenosis, and 6 cases of severe stenosis after carotid color Doppler ultrasonography. The expression levels of miR-145 and miR-210 in serum were measured using real-time fluorescence quantitative polymerase chain reaction (qPCR) technology. Results The expression levels of serum miR-145 and miR-210 in carotid artery stenosis group were significantly lower than those in non-stenosis group (p < 0.001). Multivariate Logistic regression analysis showed that smoking history, diabetes, hypertension and total cholesterol were positively correlated with the occurrence of carotid artery stenosis (p < 0.05). The expression levels of miR-145 and miR-210 were significantly negatively correlated with carotid artery stenosis (p < 0.001). In addition, patients with carotid artery stenosis and low expression levels of miR-145 and miR-210 had a greater risk of cerebral ischemia (p < 0.05). Cox regression analysis showed that the low expression of miR-145 and miR-210 were independent predictors of cerebral ischemic events. ROC analysis confirmed that miR-145 and miR-210 had good diagnostic efficacy in cerebral ischemia (p < 0.001). Conclusion The decreased expression of miR-145 and miR-210 in serum is closely related to the diagnostic significance of carotid artery stenosis, and may be used to predict the occurrence of cerebral ischemic events.