Key points In a rat model of ageing that is free of atherosclerosis or hypertension, E/A, a diagnostic measure of diastolic filling, decreases, and isovolumic relaxation time increases, indicating that both active and passive ventricular relaxation are impaired with advancing age. Resting coronary blood flow and coronary functional hyperaemia are reduced with age, and endothelium‐dependent vasodilatation declines with age in coronary resistance arterioles. Exercise training reverses age‐induced declines in diastolic and coronary microvascular function. Thus, microvascular dysfunction and inadequate coronary perfusion are likely mechanisms of diastolic dysfunction in aged rats. Exercise training, initiated at an advanced age, reverses age‐related diastolic and microvascular dysfunction; these data suggest that late‐life exercise training can be implemented to improve coronary perfusion and diastolic function in the elderly. Abstract The risk for diastolic dysfunction increases with advancing age. Regular exercise training ameliorates age‐related diastolic dysfunction; however, the underlying mechanisms have not been identified. We investigated whether (1) microvascular dysfunction contributes to the development of age‐related diastolic dysfunction, and (2) initiation of late‐life exercise training reverses age‐related diastolic and microvascular dysfunction. Young and old rats underwent 10 weeks of exercise training or remained as sedentary, cage‐controls. Isovolumic relaxation time (IVRT), early diastolic filling (E/A), myocardial performance index (MPI) and aortic stiffness (pulse wave velocity; PWV) were evaluated before and after exercise training or cage confinement. Coronary blood flow and vasodilatory responses of coronary arterioles were evaluated in all groups at the end of training. In aged sedentary rats, compared to young sedentary rats, a 42% increase in IVRT, a 64% decrease in E/A, and increased aortic stiffness (PWV: 6.36 ± 0.47 vs.4.89 ± 0.41, OSED vs. YSED, P < 0.05) was accompanied by impaired coronary blood flow at rest and during exercise. Endothelium‐dependent vasodilatation was impaired in coronary arterioles from aged rats (maximal relaxation to bradykinin: 56.4 ± 5.1% vs. 75.3 ± 5.2%, OSED vs. YSED, P < 0.05). After exercise training, IVRT, a measure of active ventricular relaxation, did not differ between old and young rats. In old rats, exercise training reversed the reduction in E/A, reduced aortic stiffness, and eliminated impairment of coronary blood flow responses and endothelium‐dependent vasodilatation. Thus, age‐related diastolic and microvascular dysfunction are reversed by late‐life exercise training. The restorative effect of exercise training on coronary microvascular function may result from improved endothelial function.