Almost 30 years have passed since the original demonstration that steroid receptors, comprising a subfamily of the nuclear receptor (NR) superfamily, exist as large (6-8S) non-DNA-binding complexes in hypotonic extracts (cytosol) of target cells; later such complexes were shown to correspond to a heterooligomer composed of receptor, heat shock (Hsp), and other proteins. Subsequently, an impressive number of studies have dealt with the composition of the "nonactive" complex, its dissociation and/or reassembly in vitro, possible functions of the non-receptor components, and their subcellular compartmentalization. While there is little dispute about the chaperoning role of some Hsps in such a complex, there is still no final proof of an association in vivo of NRs and Hsps in the nuclei of target cells, which is requisite for a direct regulatory involvement of Hsps in NR function. Here we critically review the various models that have been put forward to attribute a biological function to the NR-Hsp90 interaction, evaluate the corresponding experimental data, and integrate recent concepts originating from the structural and functional analyses of NRs.