Although cigarette smoking is a major cardiovascular risk factor, its microcirculatory effects are not well understood. The authors investigated cross-sectional and longitudinal relations between smoking and retinal microvascular caliber in an older Australian population. Retinal vessel calibers were measured from digitized retinal photographs at baseline (1992-1994; n = 3,006) and at 5-year follow-up examinations (1997-1999; n = 1,850) in the population-based Blue Mountains Eye Study. Smoking status and quantity and pack-years of smoking were defined from standardized questionnaires administered at both examinations. After adjustment for other factors, current and past smoking were associated with wider retinal venular caliber at baseline (p < 0.01). These cross-sectional findings were virtually identical at the 5-year follow-up examination (p < 0.001). Heavy smokers were more likely to have relatively large changes in venular caliber in either direction over the 5-year period (widening or narrowing more than 1 standard deviation greater than the mean change) than were nonsmokers (for continuous trend by pack-years or current amount smoked, p < 0.05). Weaker and less consistent associations between smoking and arteriolar caliber were observed. These results suggest long-term effects of smoking on venular caliber that may contribute to associations between smoking and cardiovascular disease.