Abstract Aims This study aimed to measure physical activity (PA) in participants with suspected heart failure with preserved ejection fraction (HFpEF) and assess associations between PA and participant characteristics. Methods and results Adults with presumed HFpEF were recruited and received diagnostic evaluation and clinical assessment. Physical activity was objectively measured using accelerometers over 7 days. To examine predictors of PA, a best subset analysis was used, with the optimal model defined as that with the lowest Bayesian information criterion. One hundred and twenty-four participants with presumed HFpEF who had valid accelerometer data were included in this study. Seventy-six were confirmed by a cardiologist as meeting the European Society of Cardiology diagnosis criteria for HFpEF. The median age of all participants was 80.1 years, and 47.4% were female. Patients spent most of each 24-h period at low-intensity PA and few or no durations at high-intensity PA, with lower activity for those with HFpEF. Gait speed was the best univariate correlate of activity levels (adjusted R2 0.29). The optimal model using best subsets regression included six variables and improved adjusted R2 to 0.47. In the model, lower levels of PA were associated with slower gait speed, lower levels of anxiety, higher levels of depression, past smoking history, a confirmed HFpEF diagnosis, and higher body mass index. Conclusion Participants demonstrated very low PA levels. The study has identified important patient characteristics associated with PA, which may help to identify those most in need of interventions. Notably, participants with confirmed HFpEF were more inactive than participants with other heart failure phenotypes.