Reproducible measures of refraction and visual acuity are necessary in clinical and population studies of eye diseases. Inter-observer differences in measurements of these parameters were evaluated using the Early Treatment of Diabetic Retinopathy Study (ETDRS) protocol. No clinically significant interobserver variation was found for refraction or visual acuity measurements between three observers. This study confirms that when standardized measurement procedures are used, observer differences as a source of variability in the data can be minimized.