Timely evaluation of cardioembolic stroke (CES) caused by atrial fibrillation is critical from the point of view of the possibility of prescribing effective secondary prevention with oral anticoagulants. Insular lesion is considered as a promising neuroimaging marker of CES.Objective: to analyze the role of insular cortex lesions using magnetic resonance imaging (MRI) of the brain as a potential neuroimaging marker of the pathogenetic subtype of ischemic stroke (IS).Patients and methods. 225 patients in the acute period of IS were examined. Depending on the stroke etiology, patients were divided into three groups: cryptogenic stroke (CS; n=99), CES (n=45), and non-CES (n=81). All patients underwent an MRI of the brain to analyze the insular cortex lesions. In 57 patients, foci of cerebral infarction were additionally marked manually on axial slices of diffusion-weighted MRI using the Anatomist software. The calculated MRI characteristics of foci for CES and non-CES groups were used to construct a decision tree in the WEKA 3.6 package. Echocardiographic markers of atrial cardiopathy were assessed in all patients – the left atrium (LA) emptying fraction and LA function index; in 68 patients, the concentration of serum NT-proBNP was also assessed.Results and discussion. The insula was affected in 12% of patients: most often in CES (33%), significantly less often in CS and non-CES (6 and 7.4%, respectively), without significant differences between the latter groups. The presence of insula lesion in relation to CES has a sensitivity of 33% and a specificity of 93% (p=0.002); odds ratio 6.25; 95% confidence interval 2.22–17.63. In most patients, the posterior insular cortex was involved in the pathological process. Isolated insular infarction occurred in only one patient with CES, while the involvement of the insula and adjacent zone, and the combination of insular infarction with territorial infarction, were observed more often. The group of patients with insular lesions was distinguished by the predominance of women, greater severity of stroke at admission, less deficit at discharge, larger LA diameter, lower LA emptying fraction, and functional index. CES was four times more common in the insular lesion group, while CS was two times more common in those without insular lesions. Insula involvement identifies three out of five CES patients according to the decision tree. Further analysis of the total lesion volume can locate almost all remaining patients with CES: they are characterized by the indicator >12 sm3.Conclusion. Insular lesions allow reliable differentiation of patients with CES and non-CES and can be considered a potential marker of the cardioembolic subtype of IS, which requires further investigation.