Background. Atrial fibrillation (AF) increases the risk for stroke but also for non-stroke major adverse cardiovascular events (MACE). The 2MACE score was recently proposed to predict these events. Since the interest of microRNAs (miRNAs) in cardiovascular diseases is increasing, we aimed to investigate whether miRNA levels may improve the predictive performance of the 2MACE score. Methods. We included consecutive AF patients stable on vitamin K antagonist therapy. Blood samples were drawn at baseline and plasma expression of miRNAs was assessed. During a median of 7.6 (interquartile range (IQR) 5.4–8.0) years, the occurrence of any MACE (nonfatal myocardial infarction/cardiac revascularization and cardiovascular death) was recorded. Results. We conducted a miRNA expression analysis in plasma from 19 patients with and without cardiovascular events. The miRNAs selected (miR-22-3p, miR-107, and miR-146a-5p) were later measured in 166 patients (47% male, median age 77 (IQR 70–81) years) and all were associated with a higher risk of MACE. The addition of miR-107 and miR-146a-5p to the 2MACE score significantly increased the predictive performance (c-indexes: 0.759 vs. 0.694, p = 0.004), and the model with three miRNAs also improved the predictive performance compared to the original score (c-indexes: 0.762 vs. 0.694, p = 0.012). 2MACE models with the addition of miRNAs presented higher net benefit and potential clinical usefulness. Conclusions. Higher miR-22-3p andmiR-107 and lower miR-146a-5p levels were associated with a higher risk of MACE. The addition of these miRNAs to the 2MACE score significantly increased the predictive performance for MACE, which may aid to some extent in the decision-making process about risk stratification in AF.