The genomic mode of action is believed to represent the predominant effect of a steroid hormone. Recently, however, rapidly manifesting, non-genomic effects have also been observed. These are mediated mostly by allosteric interaction of a steroid with heterologous target structures such as membrane receptors, a prototype example being the GABA A. Here we describe our studies considering two interdependent questions: (1) do steroids also interact with opioid receptors in brain? Twenty different steroids, i.e. estrogens, androgens, glucocorticoids, mineralocorticoids, gestagens and a cardiac glycoside were tested with respect to their ability to compete for in vitro binding to rat brain membranes of 3H-ligands specific for σ, μ and κ opioid receptors, respectively. Among all classes of steroids, only the estrogens were effective, all others were 20 to 100 times less effective or ineffective. The rank order among the estrogens was diethylstilbestrol > 17α-estradiol > 17α-ethinyl-estradiol > estriol > estrone > 17β-estradiol. Next potent to estrogens (although far less) were—on average as a group—glucocorticoids, followed by mineralocorticoids, androgens, gestagens and digoxin. This global as well as within-group rank order was, with rare exceptions, qualitatively equal irrespective of which radioligand was used, yet displayed the various radioligands different sensitivities with respect of being inhibited by steroids (irrespective of kind), i.e. in the order [ 3 H] naloxone ⩾ [ 3 H] DAGO ⩾ [ 3 H] DADL > [ 3 H] DPDP ⪢ [ 3 H] etorphine . The IC 50 of diethylstilbestrol for displacing [ 3H]DAGO was ≈30 μM and that of 17β-estradiol was ≈ 200 μM. (2) What are the concentrations of the major steroid hormones in the brain's extracellular fluid? We have analyzed in 56 matched (i.e. simultaneously withdrawn) peripheral serum and cerebrospinal fluid (CSF) samples (from endocrinologically grossly normal patients) the concentrations of the unconjugated steroid hormones testosterone, androstendione, dehydroepiandrosterone (DHEA), progesterone and cortisol (all being more or less lipophilic) as well as those of their hydrophilic counterparts, i.e. DHEA-sulfate, or their hydrophilic binding proteins, i.e. sex hormone binding globulin, corticosterone binding globulin, and albumin. Total (i.e. free plus protein-bound) CSF levels of all these steroids were found to be in the 0.02–2 nM range except for cortisol (≈ 20–50 nM), thus 3 to 4 orders of magnitude lower than the IC 50 of estrogens for [ 3H]DAGO (see above). These total CSF values were quite similar to the reported and calculated free serum levels of these steroids and no difference existed between those of patients with intact or with disturbed (abnormally leaky) blood-brain barrier function. Thus in vitro demonstrated low affinity (micromolar) interactions of steroids with neuronal membrane-bound receptors should be considered in view of the above found or expected concentrations of steroids in the interneuronal fluid. Whether neurosteroids can in vivo reach intrabrain concentrations of neuromodulatory efficacy remains to be seen; certain steroidal drugs, unlike peripheral steroid hormones, may do so.