The reproducibility of digital photogrammetry of the [human] optic nervehead photographs taken with the Zeiss fundus camera with Allen stereoseparator, the Donaldson fundus camera and the Zeiss fundus camera with twin-prism separator were evaluated. Reproducibility of the various quantitative factors was significantly better when the 2 types of simultaneous stereophotographs were analyzed. The Donaldson fundus camera provided the best reproducibility of the majority of geometric factors in varying-size optic cups. Larger [percentage] errors were encountered with shallower optic cups. Substantial improvements in parameter reproducibility were obtained when finer sampling intervals were used during film digitization.