Background There is evidence to suggest that microvascular disease, particularly diabetic retinopathy, plays a role in the pathogenesis of HF. However, whether changes in retinal vessel calibre predicts HF is unclear. The purpose of this study was to examine the association of retinal microvascular structure with prevalent heart failure (HF). Methods The Australian Heart Eye Study (AHES) is a cross-sectional study that surveyed 1680 participants who presented to a tertiary referral hospital for the evaluation of potential coronary artery disease by coronary angiography. Retinal vessel calibre was graded using retinal photography and participants'self-reported echocardiography-confirmed HF was obtained via an extensive medical questionnaire. Results There were 107 participants (8.1%) with prevalent self-reported HF. Persons with wider retinal arteriolar calibre (comparing highest versus lowest tertile or reference) were more likely to have prevalent HF (OR 3.5; 95% CI, 1.7-7.2) when adjusted for age and sex. After further adjustment for body mass index, hypertension, diabetes, smoking status, triglycerides and estimated glomerular filtration rate, this association remained significant (OR 4.5; 95% CI, 2.0-9.8). After further stratification, this association remained significant among participants with diabetes (OR 10.3; 95% CI, 2.7-39.3) but not in those without diabetes (OR 2.7; 95% CI, 0.9-7.5). The strength of this association was not dependent on the length of history of diabetes, or retinopathy status. There was no significant association between retinal venular calibre and prevalence of HF. Conclusions Wider retinal arteriolar diameter was significantly and independently associated with prevalent HF in participants of a cross-sectional study. This association was significant stronger among participants with diabetes compared to without diabetes. No association was found between retinal venule calibre with prevalent HF.