Human strabismus amblyopes show deficits in spatial vision that are revealed in a variety of visual tasks. In particular, they show severe deficits in their ability to encode the relative spatial phase of the sinusoidal components in a compound grating. To investigate the neural basis of strabismic amblyopia we tested the ability of monkeys with experimentally induced strabismus to encode spatial phase relationships. First, we trained them to discriminate between compound gratings (made of a fundamental sinusoid and its third harmonic) that differed only in the relative phase of their components. These monkeys exhibited a pattern of severe deficits that resemble those described in the human population of strabismic amblyopes. We conclude that these animals represent a valid model of strabismic amblyopia. Second, we show that a model that had been used to account for the performance of normal human subjects and of humans with anisometropic amblyopia fails to predict the performance of monkeys with strabismic amblyopia.