Diabetic retinopathy was assessed in a population-based study of 2708 diabetic persons in southern Wisconsin. The retinopathy levels as determined by ophthalmoscopy and by the grading of stereoscopic fundus photographs were compared in the eyes of 1949 persons. Ophthalmoscopy was performed by an ophthalmologist and a specially trained optometrist and ophthalmic technician. Consultation among the three examiners was permitted. There was exact agreement between ophthalmoscopy and grading for detecting retinopathy (none, nonproliferative, proliferative) 85.7% of the time. The kappa statistic, which corrects for chance agreement, was 0.749. There were no significant differences among the three ophthalmoscopists. Ophthalmoscopy was more likely to disagree with fundus photography grading in eyes with less severe forms of retinopathy and in patients examined early in the study. Other factors found to influence the degree of agreement were age, visual acuity, and duration of diabetes. It is concluded that with proper training ophthalmoscopy can be an acceptable alternative to fundus photography in certain situations.