Stereopsis performance was assessed in 88 optometric patients using an operant match-to-sample discrimination task involving random dot stereograms (RDSs). All normals passed the RDS test, and all constant strabismics without amblyopia, microtropes, and amblyopic strabismics failed. Only a portion of anisometropia amblyopes, intermittent strabismics, and ocular-pathology patients passed. The findings were interpreted as indicating that stereopsis with a RDS may be better predicted and explained in terms of binocular fusion and bifoveal alignment than by visual acuity.