Current models of the ocular motor system are usually presented in their most reduced form, are unilateral in architecture, and precise yoking is presumed. Although this simplifies the models, it does not accurately simulate the actual neuroanatomy and limits the models to simple, stereotyped responses. Studies of normal humans and monkeys have demonstrated striking disconjugacies in normal responses. Normal saccades may be disconjugate, or 1 eye may exhibit a dynamic overshoot. Asymmetric vergence can result in disconjugate saccades, unequal magnification spectacles cause differential saccadic gain adjustment, and saccades to unequal disparities also cause unequal saccades in the 2 eyes. In strabismus, deviated eyes typically do not mimic the movements of the fixating eye nor do their latent or congenital nystagmus waveforms duplicate those of the fixating eye. In spasmus nutans, each eye oscillates independently of the other. In achiasmatic dogs, uni-ocular saccades and uni-ocular nystagmus waveforms are seen; the same may be true in human achiasma. These data from both normals and those with abnormalities suggest that current models for ocular motor control are inadequate representations of the actual system. The inability of unilateral, yoked control (or even bilateral, yoked control) system models to duplicate the ocular motor responses of binocular mammals suggests that their ocular motor systems evolved from the bilateral, independent control systems seen in chameleons. One need only postulate a yoking overlay superimposed on two independent control systems to achieve conjugacy (bilateral, yoked, independent control) of the eyes. Abnormalities producing grossly disconjugate eye movements may then be simulated using the independent control of each eye released by a deficiency in the yoking overlay. Independent control of each eye coupled with the essential bilateral brain stem architecture implies that each individual muscle is driven by independent populations of neurons (burst cells, neural integrator cells, etc.). The agonist muscles of each eye are usually coordinated (yoked) but may function independently if the task dictates or if binocularity did not develop. Models based on the above architecture would be robust and could duplicate the many responses (both normal and abnormal) possible from the neurophysiological system.