Aims To determine 5-year incidence and progression of diabetic retinopathy in an older Australian population-based cohort. Methods During the period 1992-1994, the Blue Mountains Eye Study examined 3654 residents aged 49 þ years (82.4% of those eligible), living in two urban postcode areas, west of Sydney, Australia. Participants were subsequently invited to attend 5-year follow-up exams. After excluding 543 (14.8%) who died during the follow-up period, 2334 persons (75.0%) were re-examined during 1997-1999. The examination included a comprehensive questionnaire, blood pressure measurement, standardised refraction, Zeiss stereo retinal photographs, and estimation of fasting blood glucose. Diabetic retinopathy was graded from the retinal photographs, using the modified Early Treatment Diabetic Retinopathy Scale classification (15-step scale). Results Of participants with diabetes diagnosed at baseline, 150 were re-examined, including 139 with gradable fundus photographs. The cumulative 5-year incidence of diabetic retinopathy was 22.2% before 95% confidence interval (CI) 14.1-32.2%. Retinopathy progression (1 þ steps) was documented in 25.9% (95% CI 18.8-34.0%) of participants with retinopathy and gradable photographs at both visits; in 58.3% of these cases, a 2 þ -step progression was documented. Progression to proliferative retinopathy occurred in only 4.1% of those with retinopathy at baseline. The only baseline risk factors associated with retinopathy progression, after adjusting for age and gender, were increase in fasting blood glucose, odds ratio (OR) 1.2 (95% CI 1.1-1.4)/ mmol/l, and increase in diabetes duration, OR 2.3 (95% CI 1.0-5.3)/10 years. Conclusions These data provide 5-year cumulative incidence of diabetic retinopathy in a defined older population. Increase in diabetes duration and elevated baseline fasting blood glucose level predicted retinopathy incidence.