BACKGROUND:Retinal vessel caliber may be a novel marker of coronary heart disease (CHD) risk. However, the sex-specific effect, magnitude of association, and effect independent of traditional CHD disease risk factors remain unclear. PURPOSE:To determine the association between retinal vessel caliber and risk for CHD. DATA SOURCES:Relevant studies in any language identified through MEDLINE (1950 to June 2009) and EMBASE (1950 to June 2009) databases. STUDY SELECTION:Studies were included if they examined a general population, measured retinal vessel caliber from retinal photographs, and documented CHD risk factors and incident CHD events. DATA EXTRACTION:6 population-based prospective cohort studies provided data for individual participant meta-analysis. DATA SYNTHESIS:Proportional hazards models, adjusted for traditional CHD risk factors, were constructed for retinal vessel caliber and incident CHD in women and men. Among 22,159 participants who were free of CHD and followed for 5 to 14 years, 2219 (10.0%) incident CHD events occurred. Retinal vessel caliber changes (wider venules and narrower arterioles) were each associated with an increased risk for CHD in women (pooled multivariable-adjusted hazard ratios, 1.16 [95% CI, 1.06 to 1.26] per 20-microm increase in venular caliber and 1.17 [CI, 1.07 to 1.28] per 20-microm decrease in arteriolar caliber) but not in men (1.02 [CI, 0.94 to 1.10] per 20-microm increase in venular caliber and 1.02 [CI, 0.95 to 1.10] per 20-microm decrease in arteriolar caliber). Women without hypertension or diabetes had higher hazard ratios. LIMITATION:Error in the measurement of retinal vessel caliber and Framingham variables was not taken into account. CONCLUSION:Retinal vessel caliber changes were independently associated with an increased risk for CHD events in women.