The Baltimore Eye Survey was a population–based survey conducted from January 1985 to November 1988 among residents of east Baltimore, Maryland, who were 40 years of age or older. A total of 5,308 black subjects and white subjects received a comprehensive screening examination for glaucoma including tonometry, visual fields, stereoscopic fundus photography, and a detailed medical and ophthalmic history. Based on a definitive examination, a diagnosis of glaucoma of any type was made for 196 persons. Tonometry, cup:disc ratio, and narrowest neuroretinal rim width were evaluated for their ability to correctly classify subjects into diseased or nondiseased states. There were no cutoff values at which these variables provided a reasonable balance of sensitivity and specificity, separately or in combination. Logistic regression models were fit that included demographic and other risk factors. Sensitivities and specificities were calculated for varying cutoff levels on the distribution of predicted probabilities. There was no cutoff for which reasonable sensitivity and specificity were obtained. The authors conclude that the effectiveness of current techniques for glaucoma screening is limited. Am J Epidemiol 1991 ;134:1102–10