For orthoptists and ophthalmologists, anomalous retinal correspondence (ARC) is a reality and an important finding. But since it has not been found in animals, ARC seems to be unknown to neurophysiologists. Comparing results of different stereotests, e.g., random-dot stereograms and the two-pencil test, provides some insight into different levels of cortical binocular interaction. Patients with orthotropia and normal retinal correspondence (NRC) and even those with anisometropic amblyopia usually pass random-dot stereograms, whereas strabismic patients with ARC, even with microtropia, usually fail. Microtropic patients, however, may pass contour stereograms, and, in large esotropic angles, useful, daily-life binocular stereopsis can be found with the two-pencil test. Random-dot stereopsis suggests that normal binocular interaction must take place in or near area 17, where data processing for small dots occurs before form recognition. Anomalous correspondence most probably has its seat where the retinal topology is not exact, i.e., where the binocular receptive fields are very large and encompass the corpus callosum, such as in area 20 or 21. This new hypothesis may explain the different forms of ARC according to the clinical entities.