Cortisol levels in the serum and cerebrospinal fluid (CSF) were studied in ovariectomized, estrogen-treated monkeys during either prolonged hypercortisolemia or the more transient effects of a bolus injection of cortisol. Control (saline-treated) animals showed the expected diurnal rhythm in serum cortisol, but proportionately more cortisol was present in the CSF when serum levels were high (i.e. in the morning). Prolonged hypercortisolemia for up to 37 days was produced by either thrice daily injections of cortisol itself or single daily injections of ACTH1-24 in the morning. Both treatments produced disproportionately larger amounts of cortisol in the CSF than in the serum, and the CSF/serum cortisol ratio was increased. Furthermore, prolonged ACTH treatment caused a marked elevation in CSF cortisol in afternoon samples taken at 16.30 h compared with those at 10.00 h, in the absence of a similar change in serum cortisol levels. The relative importance of entry and clearance of cortisol in the CSF in these conditions was studied in several ways. 'Free' cortisol levels in serum (determined by equilibrium dialysis) were equal to CSF cortisol levels in control monkeys, but were less than those in the CSF of hypercortisolemic animals. Entry of cortisol into CSF after a bolus injection was rapid, but, unlike serum, CSF cortisol levels did not decline significantly over a 70-min sampling period and the delayed clearance from the CSF could account for some of the effects seen during hypercortisolemia. Neither high levels of prolactin (which is elevated together with cortisol in 'stress'), induced by giving sulpiride, nor treatment with progesterone (which is also bound by corticosteroid binding globulin) altered the distribution of cortisol between blood and CSF. The concentrations of cortisol in the CSF therefore are regulated by factors influencing both its entry and clearance from the cerebral compartment. Neural tissues sensitive to cortisol are thus exposed to levels of this hormone that are both qualitatively and quantitatively different from those expected by direct extrapolation from serum levels.