Ocular convergence is usually accompanied by excyclotorsion of the eyes. Furthermore, the magnitude of cyclotorsion is dependent on the elevation of the eyes. The reason for this excyclotorsion during convergence is not understood. Excyclotorsion could be produced by either increased activity in the inferior oblique muscle or decreased activity in the superior oblique muscle. An earlier study indicated that convergence may also be accompanied by a temporal (lateral) translation of the eye. This observation is more consistent with a relaxation of the superior oblique than contraction of the inferior oblique. This hypothesis was tested by recording the activity of 31 neurons in the trochlear nucleus, which contains the superior oblique motoneurons. This was done in alert monkeys that were trained to make both versional and vergence eye movements. In addition, the cyclotorsion associated with convergence was measured in one of these monkeys. A consistent excyclotorsion associated with convergence was observed. Trochlear unit activity decreased during convergence in all cells tested. The magnitude of this decrease was significantly greater than that seen with conjugate adduction. Furthermore, the size of the decrease varied systematically with ocular elevation in a manner that was consistent with earlier measures of cyclotorsion during convergence. These results suggest that the excyclotorsion seen during convergence, and perhaps the lateral translation of the eye, are due to a relaxation of the superior oblique muscle. This relaxation during convergence is greater than that which accompanies similar conjugate movements of the eyes. We hypothesize that this peculiar pattern of muscle innervation has a motor rather than sensory function. It is suggested that relaxation of the superior oblique muscle aids adduction by partially compensating for the failure of the lateral rectus muscle to relax sufficiently during convergence. This hypothesis is consistent with our previous observations of the activity of putative medial and lateral rectus motoneurons during symmetrical convergence. Because the superior oblique also depresses the eye, decreased trochlear activity would tend to allow the eye to move upward during convergence. To counter this tendency, one might expect to see decreased activity in extraocular motoneurons with upward on-directions and/or increased activity in those with downward on-directions. This hypothesis has not been systematically evaluated. The results of this study, together with previous reports of oculomotor and abducens activity, indicate that the patterns of extraocular motoneuron innervation for similar eye positions may be quite different, depending on the angle of convergence. Thus the adduction associated with convergence is not equivalent to the adduction associated with conjugate eye movements. These differences may result from the different premotor circuitry and connections of the conjugate and vergence oculomotor systems.