PURPOSE: To determine the reproducibility over time of visual estimates of the horizontal cup/disk ratio by trained technicians from optic disk stereophotographs. METHODS: Baseline optic disk stereophotographs are graded at entry and regraded annually in a masked fashion. The 1,636 participants in the Ocular Hypertension Treatment Study (OHTS) undergo stereoscopic optic disk photography at study entry and annually thereafter. Stereophotographs are graded independently by two technicians at the Optic Disc Reading Center. If the readers' estimates of horizontal cup/disk ratio differ by more than 0.2 disk diameters (DD), they attempt to reach a consensus; if they cannot, the horizontal cup/disk ratio is adjudicated by a glaucoma specialist. RESULTS: The percent of regradings differing by 0.2 DD or more from the estimate of horizontal cup/disk ratio made at entry was 4%, 6%, and 7%, respectively at years 1, 2, and 3. The percent differing by more than 0.2 DD was 1% or less at all years. Intraclass correlation coefficients were 0.93, 0.92, and 0.92, respectively. Estimates of horizontal cup/disk ratio from sequential full-frame photographs and simultaneous split-frame photographs appeared comparable and equally reproducible. Gradings by technicians were comparable to gradings by glaucoma specialists. CONCLUSIONS: High reproducibility between repeated gradings of baseline horizontal cup/disk ratio was achieved by trained technicians adhering to a rigorous protocol. Horizontal cup/disk ratio measurements in OHTS are sufficiently reproducible to provide information about the relationship of cup/disk ratio to the prognosis of individuals with ocular hypertension.