AIM: Fractal analysis provides a global assessment of vascular network architecture. We examined the relationship of retinal vascular fractal dimension (D(f)) with coronary heart disease (CHD) mortality. METHODS AND RESULTS: We examined the relationship of D(f) with 14-year CHD mortality in a prospective, population-based cohort of 3303 participants aged 49 years or older. D(f) was measured from digitized fundus photographs using computer-automated methods; CHD mortality was documented from Australian National Death Index records. Mean D(f) in this population was 1.441 (standard deviation, 0.024). Over 14 years, there were 468 (14.2%) CHD deaths. Participants with suboptimal D(f) (lowest and highest quartiles) had 50% higher 14-year CHD mortality than those with optimal D(f) (middle quartiles), after adjusting for age, blood pressure, and other risk factors. Among participants aged ≤ 70 years, suboptimal D(f) was associated with a nearly two-fold higher risk of CHD mortality [adjusted hazard ratio (HR) 1.89, 95% confidence interval (CI), 1.25, 2.84 for the lowest quartile and HR 1.87, CI 1.30, 2.69 for the highest quartile, compared with middle quartiles]. CONCLUSIONS: D(f) is a novel means of quantifying microvascular branching that independently predicted 14-year CHD mortality. These findings suggest that suboptimal microvascular branching may play a role in development of clinical cardiovascular disease.