The effects of bilateral lesions within and around the prepositus hypoglossi (p.h.) nuclei on the optokinetic system were studied. The pure optokinetic nystagmus (o.k.n.) was evoked by a step of velocity (60 deg/s, 30 s duration) of the surrounding. The visual-vestibular interaction was investigated by measuring the gain and phase of the vestibulo-ocular reflex (v.o.r.) as a function of frequency before and after lesion under three different conditions of testing: basic v.o.r. tested in the dark, v.o.r. tested in the light and v.o.r. suppressed by vision. The tested amplitude was +/- 20 deg. A posterior vermectomy was performed for controls in two cats. A bilateral electrolytic p.h. lesion including the rostral pole of this nucleus was added to the posterior vermectomy in three cats. A lesion similar but sparing the rostral pole of the nucleus was carried out in three other cats. In one cat a bilateral electrolytic lesion of the medial vestibular nuclei (m.v.n.) was combined with a posterior vermectomy. In two cats the medulla was cut on the mid line after a posterior vermectomy. The posterior vermectomy affected neither the optokinetic response nor the visual-vestibular interactions. In cats where p.h. lesion included its rostral pole and in the cat with m.v.n. lesion, all the tested optokinetic effects (step o.k.n., and visual-vestibular interactions) were abolished. In the three cats where p.h. lesion spared its rostral pole, the optokinetic effects were quite normal in one cat, mildly reduced in the second one, and seriously affected but not completely abolished in the third one. The surgical cut of the medulla on the mid line did not dramatically disturb the various optokinetic effects. The most marked deficit was the loss of the optokinetic after-nystagmus (o.k.a.n.). From the comparison of these results with the neuroanatomical data and with the Robinson's model concerning the optokinetic processing, it was suggested that: (a) the rostral p.h. could be the location of the o.k.n. integrator or could be an essential link on the o.k.n. pathway, (b) the posterior four-fifths of the p.h. could not be an essential relay on the o.k.n. pathway, (c) the loss of o.k.a.n. after mid-line lesion could be due to the interruption of the positive feed-back loop formed by the reciprocal inhibitory connexions between the two m.v.n.