To study the effect of hypoglycaemia on secretion rates of corticotrophin-releasing hormone (CRH), arginine vasopressin (AVP) and ACTH in a non-ruminant species, a non-surgical method was used to collect pituitary venous (PitVen) blood every 0.5 or 1 min from seven horses before and after insulin administration (0.4 U/kg i.v.). To assess the effect of PitVen cannulation on results, peripheral hormones were also measured before and after insulin in five horses without PitVen cannulae. Insulin administration lowered plasma glucose in all horses (P < 0.0001; paired t-test). Cortisol concentrations, which were similar in horses with and without PitVen cannulae before insulin, rose significantly after insulin administration in both groups. Most horses showed discomfort as glucose fell. When data from horses with and without PitVen cannulae were pooled, the peak fractional change in cortisol (Spearman's rank correlation coefficient (rs) = -0.94, P < 0.001) and the severity of hypoglycaemic symptoms (rs = -0.61, P < 0.02) were inversely ranked with the glucose nadir. In horses with PitVen cannulae, insulin administration increased secretion rates of ACTH (P < 0.0001), AVP (P < 0.0001) and CRH (P < 0.02). Increments in ACTH (rs = -0.96, P < 0.005) and CRH (rs = -0.81, P < 0.05), but not in AVP, measured during the second half-hour after insulin (i.e. the peak response), were inversely ranked with the glucose nadir. Moreover, ACTH increments were positively ranked with those in CRH (rs = 0.81, P < 0.05), but not in AVP. Nevertheless, in individual horses, minute-to-minute AVP and ACTH concentrations in PitVen blood were always correlated, whereas minute-to-minute CRH and ACTH concentrations were correlated only when glucose dropped below 3.4 mmol/l. In less hypoglycaemic horses, ACTH secretion rose despite little or no change in CRH. We suggest that in horses AVP is the primary acute signal for ACTH release both before and during hypoglycaemia; however, the increasing magnitude of ACTH increments induced by greater degrees of hypoglycaemia is determined largely by selective CRH release, which then augments corticotroph responses to AVP.