Growth hormone (GH) is secreted as pulses in vivo. To understand the signals governing this periodicity, we have established a perifusion-based model of pulsatile GH release. Male rat anterior pituitaries were dispersed and perifused with pulses of human growth hormone-releasing factor-(1--40) (GHRF), with or without a continuous or discontinuous somatostatin tonus. An experiment was composed of a 1-h base-line collection followed by four 3-h cycles; each contained single or paired 10-min infusion(s) of 3 nM GHRF. In testing the impact of somatostatin, the protocol was identical except that 0.3 nM somatostatin was added 30 min into the base-line period and then was either continued throughout the study or withdrawn during the periods of GHRF infusion. GH base lines with somatostatin were lower than vehicle base lines (P less than 0.05). GHRF pulses generated consistent peaks of GH release between 200 and 300 ng. min-1. (10(7) cells)-1, and these peaks were not altered by continuous somatostatin. In contrast, withdrawal of somatostatin during GHRF administration elicited markedly higher GH peaks (P less than 0.05) and more total GH release (P less than 0.05). This response could not be accounted for by the additive effects of GHRF and somatostatin withdrawal.