Background: Left ventricular (LV) and arterial stiffening occurs with aging, which can be prevented by lifelong exercise training. Heart failure with preserved ejection fraction (HFpEF) is a disease of the elderly with stiff LV and arteries. However, it is unknown whether exercise training in HFpEF could improve beat-to-beat ‘dynamic’ Staring mechanism, which reflects the interaction between LV and arterial stiffness or LV-arterial coupling. Methods: Seven patients with HFpEF (age 75 ± 6 yrs) and 9 healthy sedentary controls (age 71 ± 3 yrs) were enrolled and participated in 1 year of endurance exercise training. Maximal oxygen uptake (VO 2 max) was measured during incremental treadmill exercise. All subjects underwent right heart catheterization; beat-to-beat LV end-diastolic pressure (LVEDP; estimated from pulmonary arterial diastolic pressure) and stroke volume (SV) index (from pulse contour analysis) were obtained during controlled breathing (0.2 Hz, 12 breaths/min) before and after 1 year of training. Spectral transfer function analysis was used as an index of the dynamic Starling mechanism (Shibata JAP2011), and effects of exercise training on the ‘dynamic’ Starling mechanism in HFpEF and healthy controls were compared. By the end of the 1 year of training, the exercise duration was about 200 minutes/wk. Results: VO 2 max was significantly larger in healthy controls at baseline. After exercise training, VO 2 max was unaffected in HFpEF, but increased in healthy controls (HFpEF: 12.6 ± 3.6 vs. 12.8 ± 3.3 ml·kg -1 ·min -1 , p = 0.71, healthy control: 22.8 ± 3.4 vs. 27.2 vs. 4.3 ml·kg -1 ·min -1 , p < 0.01, ANOVA p < 0.01). The ‘dynamic’ Starling mechanism (transfer function gain between LVEDP and the SV index) at baseline was impaired in HFpEF patients compared to the healthy control (HFpEF 0.26 ± 0.14 vs. 0.40 ± 0.09 ml·ml -2 ·mmHg -1 , p = 0.026). Transfer function gain was improved by 65% after exercise training in healthy controls (p = 0.035), while gain in HFpEF was not improved by exercise training (0.26 ± 0.14 vs. 0.29 ± 0.17 ml·ml -2 ·mmHg -1 , p = 0.78). Conclusion: A year of exercise failed to improve beat-to-beat LV-arterial coupling in HEpEF, while it was improved in healthy controls. These results may partly explain why no improvement was observed in VO 2 max in HFpEF.