Adrenalectomized rats displayed a deficiency in retention of an immobility response acquired during an initial 15-min forced swimming procedure (Porsolt swimming test) and measured 24 h later in a 5-min retest session. The deficit could be restored dose dependently with the glucocorticoid dexamethasone (µg range) administered 15 min after the initial test. The antiglucocorticoid RU 38486 administered subcutaneously (1 and 10 mg/kg) inhibited the dexamethasone effect and caused a parallel shift in the dose-response curve of dexamethasone. Intracerebroventricular administration of RU 38486 to intact rats immediately before the initial test attenuated retention of acquired immobility over a 100,000-fold lower dose range (ng) and increased the plasma corticosterone level. Local administration of 1 ng RU 38486 in the dentate gyrus of the hippocampus also diminished the percentage immobility, but did not influence the adrenocortical response. Injections of RU 38486 in parafascicular and paraventricular nucleus were ineffective on behaviour. In the latter nucleus the antiglucocorticoid disinhibited the activity of the hypothalamus-pituitary-adrenal axis. Intracerebroventricular pretreatment with promegestone did not interfere with RU 38486 action, ruling out involvement of its antiprogestin properties. Intracerebroventricular or subcutaneous treatment of intact rats with the antimineralocorticoid RU 28318 was not effective. Finally, adrenalectomized rats replaced with corticosterone delivered via subcutaneously implanted 100-mg corticosterone pellets showed normal behavioural performance, while a 25-mg implant did not. The present study with local infusions of RU 38486 indicates that glucocorticoid feedback via type 2 receptors exerts a long-term influence on behaviour in the hippocampus and controls the activity of the hypothalamus-pituitary-adrenal axis in the paraventricular nucleus.