Change in optic disc cupping was evaluated in a 4-year follow-up of a well defined cohort of people with diabetes mellitus. Cup-to-disc ratios were computed for both vertical and horizontal diameters of each eye at the baseline and 4-year follow-up examinations. Graders were masked as to the identity of participants and to the dates of the photographs. Increases of at least 0.1 between baseline and follow-up were used as clinically significant change in the ratios. None of the following factors at baseline were consistent predictors of such a change: intraocular pressure, age, duration of diabetes, hypertension or severity of diabetic retinopathy. People who developed proliferative retinopathy by the follow-up examination were not more likely to have such an increase in ratio at the follow-up. We conclude that clinically significant increases in cup-to-disc ratio cannot be consistently predicted in people with diabetes from the risk factors evaluated with the grading system used in this study.