To characterize the vergence signal carried by the medial longitudinal fasciculus (MLF), it was subjected to reversible blockade by small injections of 10% lidocaine hydrochloride. The effects of these blockades on both conjugate and vergence eye movements were studied. With this procedure, experimentally induced internuclear ophthalmoplegia (INO) and its effects on conjugate eye movements could be studied acutely, without possible contamination from long-term oculomotor adaptation. In the eye contralateral to the MLF blockade, saccadic and horizontal smooth-pursuit eye movements were normal. Horizontal abducting nystagmus, often seen in patients with INO, was not observed in this eye. As previously reported for INO, profound oculomotor deficits were seen in the eye ipsilateral to the MLF blockade. During maximal blockade, adducting saccades and horizontal smooth-pursuit movements in this eye did not cross the midline. Adducting saccades were reduced in amplitude and peak velocity and showed significantly increased durations. Abducting saccades, which were slightly hypometric, displayed a marked postsaccadic centripetal drift. The eye ipsilateral to the blockade displayed a pronounced, upward, slow drift, whereas the eye contralateral to the blockade showed virtually no drift. Furthermore, although vertical saccades to visual targets remained essentially conjugate, the size of the resetting quick phases in each eye was related to the amplitude of the slow phase movement in that eye. Thus the eye on the affected side displayed large quick phases, whereas the eye on the unaffected side showed only slight movements. On occasion, unilateral downbeating nystagmus was seen. This strongly suggests that the vertical saccade generators for the two eyes can act independently. The effect of MLF blockade on the vergence gain of the eye on the affected side was investigated. As a measure of open-loop vergence gain, the relationship of accommodative convergence to accommodation (AC/A) was measured before, during, and after reversible lidocaine block of the MLF. After taking conjugate deficits into account, the net vergence signal to the eye ipsilateral to the injection was found to increase significantly during the reversible blockade. The most parsimonious explanation for this increased vergence signal is suggested by the accompanying single-unit study. This study showed that abducens internuclear neurons, whose axons course in the MLF, provide medial rectus motoneurons with an appropriate horizontal conjugate eye position signal but an inappropriate vergence signal. Ordinarily, this incorrect vergence signal is overcome by another, more potent, vergence signal impinging on the medial rectus motoneurons. This stronger signal is revealed when the inappropriate MLF vergence signal is reduced by lidocaine blockade.