To assess the association between atrial fibrillation (AF) and mortality, and also the association between resting heart rate (HR) and mortality in both sinus rhythm (SR) and AF in patients with heart failure with preserved ejection fraction (HFpEF). A total of 9090 patients with HFpEF (ejection fraction ≥ 50%) were included from the Swedish Heart Failure registry; 4296 (47%) had SR and 4794 (53%) had AF. Patients with AF were older (80.3 vs. 75.0 years) and more symptomatic compared with patients in SR. The outcome measure was all-cause mortality. The adjusted hazard ratio (95% confidence interval) for AF vs. SR was 1.21 (1.11-1.32). Compared with HR ≤ 60 b.p.m., the adjusted hazard ratios (95% confidence interval) were in SR: 1.06 (0.92-1.21) for HR 61-70 b.p.m., 1.30 (1.12-1.52) for HR 71-80 b.p.m., 1.27 (1.07-1.51) for HR 81-90 b.p.m., and 1.77 (1.45-2.17) for HR > 90 b.p.m. Due to non-proportional hazards in AF, hazard ratios were estimated in three time periods. Compared with HR ≤ 60 b.p.m., the adjusted hazard ratios (95% confidence interval) were in AF: 1.30 (1.07-1.57), 1.07 (0.83-1.39), and 1.01 (0.70-1.48) for HR 61-70 b.p.m., 1.35 (1.12-1.62), 0.99 (0.77-1.27), and 0.96 (0.66-1.40) for HR 71-80 b.p.m., 1.41 (1.16-1.73), 1.01 (0.76-1.36), and 0.79 (0.51-1.22) for HR 81-90 b.p.m., and 1.78 (1.46-2.17), 1.08 (0.80-1.46), and 0.73 (0.46-1.17) for HR > 90 b.p.m., during 0-2, 2-4, and 4-6 years of follow-up, respectively. In a large and unselected cohort of patients with HFpEF, AF was independently associated with all-cause mortality. A higher HR was associated with increased mortality in SR. In AF, the effect of a higher HR on mortality was only present during the first years of follow-up, with convergence in outcomes according to baseline HR groups over long-term follow-up.