OBJECTIVE: To compare the diagnostic accuracy of the Moorfields regression classification (MRC) and subjective optic disc evaluation in discriminating early to moderate glaucomatous from nonglaucomatous eyes. DESIGN: Cross-sectional observational study. PARTICIPANTS: Two hundred thirty-three patients with glaucoma and 216 normal subjects were included in the analysis. Racial groups were defined by self-description. METHODS: All subjects underwent confocal scanning laser ophthalmoscopy, stereophotography, and standard perimetry. Glaucoma was defined by visual field defect alone and confirmed with a second visual field test. Stereo photographs were graded as either normal or glaucomatous appearing in a masked fashion by 2 independent graders and adjudicated by a third grader in cases of disagreement. Mean disc area was compared between patients correctly and incorrectly diagnosed with either technique. MAIN OUTCOME MEASURES: Sensitivity and specificity of MRC and subjective evaluation of stereophotographs in the detection of glaucomatous visual field loss. RESULTS: With the MRC, the sensitivity and specificity were higher using the 95% cutoff than using the 99.9% cutoff. Classification based on subjective photo assessment had a greater agreement with the diagnosis of glaucoma than the MRC for blacks (MRC, sensitivity = 62.5%, specificity = 93.2%; Photo, sensitivity = 76.5%, specificity = 91.5%) and whites (MRC, sensitivity = 67.0%, specificity = 92.2%; photo, sensitivity = 78.4%, specificity = 91.9%). Disc area was significantly larger in patients incorrectly diagnosed with the MRC (P = 0.0289). CONCLUSIONS: Subjective optic disc grading by glaucoma specialists outperformed the MRC with the HRT II in both black and white subjects. Both subjective and objective diagnostic methods were associated with similar sensitivity and specificity between racial groups. The MRC was more likely to provide an incorrect diagnosis in subjects with larger optic discs.