A cytoarchitectonic scheme for the rat paraventricular nucleus has been proposed utilizing the retrograde transport of horseradish peroxidase, immunocytochemical localization of neurophysin-positive cells and fibers, Golgi-like impregnation of neurons, and observations of cell clustering as viewed in Nissl-stained material. Beginning rostrally, it was observed that the anterior commissural nucleus, a cell group sometimes claimed to be part of the paraventricular nucleus, contained many magnocellular perikarya which:

• 1.
  (1)projected to the neurohypophysis but apparently not to the brainstem or spinal cord;
• 2.
  (2) contained neurophysin;
• 3.
  (3) had few (one or two) dendrites, most of which projected toward the third ventricle; and
• 4.
  (4) were actually separated from the paraventricular nucleus by 300–400 μm. Proceeding caudally, the medial and lateral magnocellular portions of the paraventricular nucleus exhibited mostly the same characteristics as listed for the anterior commissural nucleus, except that the cells were slightly larger and more closely packed; these latter two form the conventional paraventricular neurosecretory nucleus.

Cells of the dorsomedial cap and extreme posterior paraventricular nucleus:

• 1.
  (1) did not appear to project to the neurohypophysis, but did connect with the brainstem and spinal cord; and
• 2.
  (2) often contained neurophysin.

In addition, large cells of the extreme posterior subnucleus:

• 1.
  (1) were more fusiform than neurons in the other three regions;
• 2.
  (2) possessed more (two or three) and longer dendrites than cells of the other divisions; and
• 3.
  (3) were loosely packed.

Although some zones of overlap were noted, the results suggest that neurophysin-containing cells of the paraventricular nucleus are relatively segregated into neurohypophysial-projecting and brainstem/ spinal cord-projecting groups. This topographic separation and the morphological differentiation of these putative subnuclei suggest that these subdivisions may be controlled independently and that release of hormone from the neurohypophysis does not necessarily imply a coincidental release of hormone into other extrahypothalamic sites.