Objective. To study the effect of the level of matrix metalloproteinase type 9 (MMP-9) in the blood from the coronary sinus on the effectiveness of radiofrequency (RFA) isolation of the pulmonary vein in patients with paroxysmal atrial fibrillation (AF). Material and methods. A single-center randomized prospective study. We selected 101 patients with paroxysmal AF who met the inclusion criteria. Before RFA, blood was taken from the coronary sinus to assess the level of MMP-9. Intraoperatively, before performing RFA, blood was taken from the coronary sinus in all patients to assess the level of MMP-9. After that, all patients underwent antral isolation of the pulmonary veins according to the standard technique with the achievement of the entry and exit block for each pulmonary vein using the CARTO 3 navigation system. The content of MMP-9 was determined by the test system Bender MedSystems GmbH (Austria). Patients were monitored remotely by telephone. In case of episodes of arrhythmia, the patient was advised to provide an ECG/HM-ECG with arrhythmia. At the end of the observation period, the patients were divided into two groups: in the 1st group, patients with AF recurrence, in the 2nd group, without relapse. The groups were compared with each other. Results. In a univariate regression analysis, statistically significant differences were obtained for the following indicators: the duration of the anamnesis of AF (p=0.048), MMP-9 (p=0.001), ESR (p=0.043), AST (p=0.026), EF (p=0.034), fibrinogen (p=0.015), creatinine (p=0.041). Based on the results, we built a multivariate logistic regression model to predict arrhythmia recurrence after RFA. As a result, with an increase in MMP-9 in the blood from the coronary sinus by 1 ng/ml, the chance of atrial fibrillation recurrence increased by 0.9%. Each month of history of AF increased the chance of recurrence by 1.4%. The influence of MMP-9 on the frequency of AF relapses after RFA of pulmonary vein ostia isolation is confirmed by the ROC curve, the area under the curve was 0.833, which indicates a pronounced relationship between MMP-9 concentration and relapse after RFA. We used neural networks to identify the most significant predictors. Based on the analysis of the importance of predictors by a multilayer perceptron, the level of MMP-9 is determined as the most significant factor influencing the chance of recurrence after RFA. The normalized importance of MMP-9 as a predictor is 100%. Conclusion. In our series of observations, an increase in the level of MMP-9 in the blood of the coronary sinus was detected in 78% of patients. An increase in the level of MMP-9 by 1 ng/ml increases the risk of AF recurrence after RFA of pulmonary vein isolation by 0.9%. Each month of history of AF increases the risk of recurrence after RFA by 1%. As a result of the analysis of the importance of predictors by a multilayer perceptron, in the study group of patients, MMP-9 is determined as the most significant factor in the recurrence of atrial fibrillation after RFA.