The mesencephalic interstitial nucleus of Cajal (iC) is considered the neural integrator for vertical and torsional eye movements and has also been proposed to be involved in saccade generation. The aim of this study was to elucidate the function of iC in neural integration of different types of eye movements and to distinguish eye movement deficits due to iC impairment from that of the immediately adjacent rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF). We addressed the following questions: (1) According to the neural integrator hypothesis, all eye movements including the saccadic system and the vestibulo-ocular reflex (VOR) share a common neural integrator. Do iC lesions impair gaze-holding function for vertical and torsional eye positions and the torsional and vertical VOR gain to a similar degree? (2) What are the dynamic properties of vertical and torsional eye movements deficits after iC lesions, e.g., the specificity of torsional and vertical nystagmus? (3) Is iC involved in saccade generation? We performed 13 uni- and three bilateral iC inactivations by muscimol microinjections in four alert monkeys. Three-dimensional eye movements were studied under head-stationary conditions during vertical and torsional VOR. Under static conditions, unilateral iC injections evoked a shift of Listing's plane to the contralesional side (up to 20 degrees), which increased (ipsilesional ear down) or decreased (ipsilesional ear up) by additional static vestibular stimulation in the roll plane, i.e., ocular counterroll was preserved. The monkeys showed a spontaneous torsional nystagmus with a profound downbeat component. The fast phases of torsional nystagmus always beat toward the lesion side (ipsilesional). Pronounced gaze-holding deficit for torsional and vertical eye positions (neural integrator failure) was reflected by the reduction of time constants of the exponential decay of the slow phase to 330-370 ms. Whereas the vertical oculomotor range was profoundly decreased (up to 50%) and vertical saccades were reduced in amplitude, saccade velocity remained normal and horizontal eye movements were not affected. Bilateral iC injections reduced the shift of Listing's plane caused by unilateral injections, i.e., back toward the plane of zero torsion. Torsional nystagmus reversed its direction and ceased, whereas vertical nystagmus persisted. In contrast to unilateral injection, there was additional upbeating nystagmus. Time constants of the position integrator of the gaze-holding system did not differ between unilateral and bilateral injections. The range of stable vertical eye positions and saccade amplitude was smaller when compared with unilateral injections, but the main sequence remained normal. Dynamic vestibular stimulation after unilateral iC injections had virtually no effect on torsional and vertical VOR gain and phase at the same time when time constants already indicated severe integrator failure. Torsional VOR elicited a constant slow-phase velocity offset up to 30 degrees toward the contralesional side, i.e., in the opposite direction to spontaneous torsional nystagmus. Likewise, vertical VOR showed a velocity offset in an upward direction, i.e., opposite to the spontaneous downbeat nystagmus. Contralesional torsional and upward vertical quick phases were missing or severely reduced in amplitude but showed normal velocity. In contrast, bilateral iC injections reduced the gain of the torsional and vertical VOR by 50% and caused a phase lead of 10-20 degrees (eye compared with head velocity). We propose that the slow-phase velocity offset during torsional and vertical VOR reflects a vestibular imbalance. It therefore appears likely that the vertical and torsional nystagmus after iC lesions is not only caused by a neural integrator failure but also by a vestibular imbalance. Unilateral iC injections have clearly differential effects on the VOR and the gaze-holding function. (ABSTRACT TRUNCATED)