Abstract Background In heart failure (HF), levels of NT-proBNP are influenced by the presence of concomitant atrial fibrillation (AF), making it difficult to distinguish between HF versus AF in patients with raised NT-proBNP. It is unknown whether levels of GDF-15 are also influenced by AF in patients with HF. In this study we compared the plasma levels of NT-proBNP versus GDF-15 in patients with HF in AF versus sinus rhythm (SR). Methods In a post hoc analysis of the index cohort of BIOSTAT-CHF (n = 2516), we studied patients with HF categorized into three groups: (1) AF at baseline (n = 733), (2) SR at baseline with a history of AF (n = 183), and (3) SR at baseline and no history of AF (n = 1025). The findings were validated in the validation cohort of BIOSTAT-CHF (n = 1738). Results Plasma NT-proBNP levels of patients who had AF at baseline were higher than those of patients in SR (both with and without a history of AF), even after multivariable adjustment (3417 [25th–75th percentile 1897–6486] versus 1788 [682–3870], adjusted p < 0.001, versus 2231 pg/mL [902–5270], adjusted p < 0.001). In contrast, after adjusting for clinical confounders, the levels of GDF-15 were comparable between the three groups (3179 [2062–5253] versus 2545 [1686–4337], adjusted p = 0.36, versus 2294 [1471–3855] pg/mL, adjusted p = 0.08). Similar patterns of both NT-proBNP and GDF-15 were found in the validation cohort. Conclusion These data show that in patients with HF, NT-proBNP is significantly influenced by underlying AF at time of measurement and not by previous episodes of AF, whereas the levels of GDF-15 are not influenced by the presence of AF. Therefore, GDF-15 might have additive value combined with NT-proBNP in the assessment of patients with HF and concomitant AF. Graphic abstract