To characterize the short-term ACTH secretion pattern and to investigate factors regulating it, pituitary venous (PV) blood was collected using our nonsurgical method from 8 unperturbed horses every 20 or 30 s for approximately 1 h. In all but 1 horse, sampling occurred during the broad circadian maximum in plasma cortisol concentrations. Concentrations of corticotropin-releasing hormone (CRH; n = 7 horses), arginine vasopressin (AVP), ACTH and cortisol were measured by radioimmunoassay. In all horses, CRH, AVP and ACTH secretion patterns appeared irregular in time and amplitude. The mean (±SEM) numbers of peaks per hour detected by the cluster program were 2.8 ± 1.2, 10.1 ± 1.9 and 10.2 ± 1.4for CRH, AVP and ACTH, respectively. However, when 2- and 5-min sampling frequencies were simulated by meaning consecutive values, significantly fewer peaks were detected in each hormone. There was no correlation between the prevailing cortisol concentration and peak frequencies of CRH, AVP or ACTH. Secretion patterns of ACTH and AVP were closely related in all horses as assessed by cross correlation analysis and coincidence of peaks, although the ratio between PV ACTH and AVP concentrations fluctuated markedly within each horse. In contrast, the relationship between CRH and ACTH secretion was variable. Bivariate spectral analysis showed only a modest degree of underlying periodicity in CRH, AVP and ACTH secretion during the very short term studied. Nevertheless, distinct peaks exceeding the 95% confidence limits of white noise were observed at periods between 2 and 30 min in 5 of 7 CRH, 6 of 8 AVP and 5 of 8 ACTH spectra. Furthermore, the slope of the regression line through each spectrum did not become indistinguishable from zero, i.e. the flat white noise continuum, until mean (±SEM) periods of 2.6 ± 0.8, 1.6 ± 0.2, and 2.0 ± 0.2 min, for CRH, AVP and ACTH spectra, respectively. At the ACTH spectral maximum, the coherence coefficient, which is analogous to the squared correlation coefficient, exceeded 0.5 in comparisons of all ACTH and AVP spectra and of 5 of 7 ACTH and CRH spectra. We conclude that (1) rapid sampling is needed to define short-term CRH, AVP and ACTH secretion patterns accurately: peaks are missed when 2- or 5-min collection regimens are used, and (2) the major components in the ACTH micropulsatile secretion pattern in the horse are regulated by secretagogues. The closeness and consistency of the relationship between AVP and ACTH secretion suggest that AVP may be the primary signal for ACTH micropulses in the unperturbe