<jats:sec><jats:title>Aims</jats:title><jats:p>We aimed to derive and validate clinically useful clusters of patients with heart failure with preserved ejection fraction (HFpEF; left ventricular ejection fraction ≥50%).</jats:p></jats:sec><jats:sec><jats:title>Methods and results</jats:title><jats:p>We derived a cluster model from 6909 HFpEF patients from the Swedish Heart Failure Registry (SwedeHF) and externally validated this in 2153 patients from the Chronic Heart Failure ESC‐guideline based Cardiology practice Quality project (CHECK‐HF) registry. In SwedeHF, the median age was 80 [interquartile range 72–86] years, 52% of patients were female and most frequent comorbidities were hypertension (82%), atrial fibrillation (68%), and ischaemic heart disease (48%). Latent class analysis identified five distinct clusters: cluster 1 (10% of patients) were young patients with a low comorbidity burden and the highest proportion of implantable devices; cluster 2 (30%) patients had atrial fibrillation, hypertension without diabetes; cluster 3 (25%) patients were the oldest with many cardiovascular comorbidities and hypertension; cluster 4 (15%) patients had obesity, diabetes and hypertension; and cluster 5 (20%) patients were older with ischaemic heart disease, hypertension and renal failure and were most frequently prescribed diuretics. The clusters were reproduced in the CHECK‐HF cohort. Patients in cluster 1 had the best prognosis, while patients in clusters 3 and 5 had the worst age‐ and sex‐adjusted prognosis.</jats:p></jats:sec><jats:sec><jats:title>Conclusions</jats:title><jats:p>Five distinct clusters of HFpEF patients were identified that differed in clinical characteristics, heart failure drug therapy and prognosis. These results confirm the heterogeneity of HFpEF and form a basis for tailoring trial design to individualized drug therapy in HFpEF patients.</jats:p></jats:sec>