Abstract Background NT-proBNP-levels are often elevated in patients with atrial fibrillation (AF), but their clinical use and interpretation for predicting subsequent cardiovascular events has not been explored in an all-comers setting. Aim To determine whether NT-proBNP levels are associated with the risk of subsequent heart failure (CHF) and/or death in patients with AF and its relation to prior CHF diagnosis Method All patients with new-onset AF (defined by ICD-code I48) who had an NT-proBNP-level within +/- 14 days of the date of the AF date measured in a regional laboratory repository between 2000-2021 were included in the analysis cohort. Patients were categorized into NT-proBNP level (pg/mL) <1000, 1000-1999, ≥2000. If several NT-proBNP-levels were measured, the maximal value was used. The association between NT-proBNP level and the outcome CHF hospitalization and/or death within 3 years of diagnosis was assessed in Cox regression analyses adjusted for baseline characteristic and medications. Since there was an interaction (p<0.001) between Nt-proBNP-level and prior/new-onset CHF at time of AF diagnosis and no known CHF, two separate analyses were performed in patients with or new-onset CHF and those without prior CHF. Results There were 10559 patients with AF and no CHF, and 11 257 with AF and CHF. There were 14% of patients with new-onset CHF at time of new-onset AF diagnosis. These were grouped together with patients with known CHF. Patients with no prior CHF were younger (76 years (95% CI 68-83) vs 80 (72-87), p<0.001), and healthier with less often diabetes (18.4% vs 24.5%, p<0.001), prior myocardial infarction (12.0% vs 24.3%), a lower CHA2DS2-VASc score (3.0 vs 5.0, p<0.001) and a lower median NT-proBNP level (1650 pg/mL (95% CI 702-3549) vs 4240 pg/mL (2160-8860), p<0.001). Among patients with AF and no prior CHF, the incidence rate in the lowest NT-proBNP level was low, but increased with higher levels (Table). Adjusting for covariates and medications, the risk of both CHF hospitalization and CHF/death compared to the reference category NTproBNP< 1000 pg/mL was nearly 2 and 3 times higher among those who with initial NT-proBNP level 1000-1999 and ≥ 2000. Among patients with AF and known CHF, the incidence rate of CHF hospitalization or CHF hospitalization/death at the same NT-proBNP-strata was several times higher compared to patients without known CHF. Adjusting for covariates, the risk increased for CHF hospitalization or CHF/hospitalization similarly for each strata as for patients without known CHF. AF patients without known HF but with the highest elevation of NTproBNP-level had a risk of subsequent hospitalization of heart failure which was higher than those with known heart failure and lowest NTproBNP-level. Conclusion A higher NTproBNP-level in patients with AF identifies patients at higher risk of CHF hospitalization and/or death in both patients with no and prior CHF.Table