Abstract Background Incident atrial fibrillation (AF) marks an adverse shift in the prognosis of patients with aortic stenosis (AS). Identifying risk factors for AF is therefore of paramount importance for timely intervention in patients with AS. In patients without AS, brain natriuretic peptides (BNP) is a well-established biomarker for left ventricular pressure overload on the pathway to heart failure and atrial fibrillation. However, a potential role of NT-proBNP to predict risk of new-onset AF in asymptomatic patients with mild to moderate AS is not well studied. Methods We included 1,434 patients with mild to moderate AS from the SEAS Study (Simvastatin and Ezetimibe in Aortic Stenosis) without AF or clinically overt heart failure at baseline. The primary endpoint for this substudy was time to incident AF, as determined by the first annual in-study 12-lead ECG with AF. Multivariable Cox model were adjusted for other important predictors of incident AF as selected by Bayesian statistics. Fine and Gray competing risk regression was used to evaluate the influence of all-cause mortality on selected predictor variables of incident AF. Results During a median follow-up of 4.3 years (range 0.1–6.9 years), incident AF occurred in 114 (6.1%) patients (13.8 per 1,000 person-years of follow-up), who at baseline were older (69±10 vs. 67±10 years, p<0.001), had larger systolic left atrial diameter (46±24 vs. 34±18 mm, p<0.001) and higher NT-proBNP level (286 [132; 613] vs. 154 [82; 297] pg/ml, p<0.001); but same left ventricular ejection fraction (66±6 mm vs. 67±6, p=0.4). In multivariable Cox regression, adjusted for age, circumferential end-systolic stress, left atrial volume and ECG PR interval, Ln(NT-proBNP) was associated with higher risk of new-onset AF (HR: 1.9 [95% CI: 1.6–2.3], p<0.001). Similar results were found when using Fine and Gray estimates with all-cause mortality (HR: 2.0 [95% CI: 1.7–2.4], p<0.001 (Figure, panel A). NT-proBNP level added incremental predictive information on incident AF over the other important, as selected by Bayesian statistics, predictor variables (C-index 0.81, p<0.001, Figure, panel B). There was no interaction with aortic valve area (p>0.05). Figure 1 Conclusions In patients with asymptomatic aortic stenosis and sinus rhythm at baseline, NT-proBNP levels were significantly higher in patients who subsequently developed AF. NT-proBNP significantly improved prognostic information of incident AF over other important predictor variables. This supports the notion that incident AF is a marker of left ventricular pressure overload and possibly a novel marker of timely intervention with aortic valve replacement.