Microdissected rat median eminences (wet weight, 110 micrograms) were electrically stimulated in vitro in the presence of bacitracin and bovine serum albumin. Incubation media collected at 20-min intervals, were assayed for corticotropin-releasing factor (CRF) with dispersed pituitary cells and for arginine vasopressin (AVP) by a specific RIA. CRF activity of the media or of synthetic AVP was plotted as function of AVP content. Dose (AVP)-response (CRF) curves for media were not significantly different from those of synthetic AVP. Extracts of incubated median eminences had 3 times more CRF activity then the AVP contained within these extracts, showing that CRF distinct from AVP ws contained within the median eminences but had not been released into or had been degraded within the media. In the presence of 1 mM ascorbic acid, AVP release was not significantly different from AVP release in the absence of ascorbic acid, but accounted for only 49-67% of released CRF activity, suggesting that a CRF distinct from AVP had been released into or protected within the medium. At an AVP level similar to that found in hypophysial portal plasma, the total CRF activity of median eminence media were not significantly different in the presence or absence of 0.13 microM haloperidol or 1.3 microM propranolol in ascorbic acid-rich media, but were 12-17% (P less than 0.05) lower in the presence of 0.35 microM phentolamine. The results indicate that CRFs, distinct from AVP, released from the median eminence in vitro can be detected only when ascorbic acid is present in the medium, and furthermore, that AVP and some alpha-adrenergic agonist account for at least 61-84% of released CRF activity.