In order to test the hypothesis that glaucomatous and nonglaucomatous optic disc cupping can be distinguished ophthalmoscopically, three ophthalmologists experienced in assessing optic dics were asked to view fundus stereophotographs as "unknowns". Of 29 eyes with nonglaucomatous optic atrophy, 13 (44%) were misdiagnosed as showing glaucoma by at least one observer. Of four optic disc features specifically analyzed, neuroretinal rim pallor proved to be 94% specific for nonglaucomatous atrophy, while focal or diffuse obliteration of the neuroretinal rim was 87% specific for glaucoma. These two signs are the most useful in making the distinction. Thinning of the rim was more common in glaucoma than in nonglaucomatous atrophy, but was only 47% specific for glaucoma. Laminar dots were present in both types of excavation.