Previous work has shown neurons just dorsal and lateral to the oculomotor nucleus that increase their firing rate with increases in the angle of ocular convergence. It has been suggested that the output of these midbrain near response cells might provide the vergence command needed by the medial rectus motoneurons. However, lens accommodation ordinarily accompanies convergence, and a subsequent study showed that only about one-half of these midbrain near response cells carried a signal related exclusively to vergence. One hypothesis suggested by this finding is that this subgroup of neurons might have a unique role in providing a 'pure' vergence signal to the medial rectus motoneurons. The 17 near response cells that could be antidromically activated from the oculomotor nucleus presumably provide vergence signals to the medial rectus motoneurons. Although all had positive vergence coefficients, only four of these cells carried signals that were related exclusively to vergence. Four other identified cells had negative accommodation coefficients, and the remainder (9 cells) had positive accommodation coefficients. The negative and positive accommodation coefficients for these cells tended to cancel, so that the overall signal carried by this subset of antidromically identified cells was related predominantly to vergence. These findings indicate that near response cells that provide the input to the medial rectus motoneurons for vergence movements are a heterogeneous group. Most, if not all, are driven by both the blur and disparity controllers but differ in their accommodative and vergence coefficients. The most parsimonious explanation for this distribution of coefficients is that individual neurons differ in the relative strengths of input they receive from the blur and disparity controllers. The hypothesis that relates the vergence and accommodation coefficients to the strengths of these signals is a simple extension of the previously described dual interaction model of accommodation and vergence. According to this extended model, near response cells that carry a signal related exclusively to vergence are simply cells for which the ratio of blur controller input to disparity controller input happens to match that of the system as a whole. Near response cells that have negative accommodation coefficients are those that receive a blur controller input that is somewhat less than the average, whereas cells with a positive accommodation coefficient have a blur controller input that is greater than the system average. The overall average of blur controller input to disparity controller input for the population of cells projecting to the extraocular motoneurons determines the overall accommodative convergence to accommodation (AC/A) ratio of the system. Because one-half of the blur controller input to the vergence system will be above the mean and one-half will be below it, the model predicts that a representative sample of near response cells in the vergence output path will have both positive and negative accommodation coefficients that will sum to zero. Data from antidromically identified cells are consistent with this prediction. Similar arguments can be made regarding near response neurons that may provide an input to the neurons controlling lens accommodation. The finding of midbrain near response cells with signals related predominantly to accommodation but with either positive or negative vergence coefficients is consistent with this hypothesis. A cell in the accommodation output path that has an input from the disparity controller that is equal to the overall CA/C (convergence accommodation to convergence) ratio would have a positive accommodation coefficient and a zero vergence coefficient. A cell that received a weaker input from the disparity controller would have a positive accommodation coefficient and a negative vergence coefficient. It should be noted that near response cells that are in the accommodative output path and receive a stronger input from the disparity controller than the blur controller would have both positive accommodation and vergence coefficients. Therefore it is not possible to determine, by measuring accommodation and vergence coefficients alone, to which subsystem a neuron belongs. The results of this study refute the hypothesis that a special subset of 'pure vergence' near response cells provide the vergence input to medial rectus motoneurons, or that such a subset exists as a distinct category of cells. Although no cells that were antidromically identified as being in the vergence subsystem were found to carry a signal related exclusively to accommodation, an analysis of the model indicates that this is possible for some cells.