The dependence of hormone binding to glucocorticoid receptors (GRs) on cellular ATP levels led us to propose that GRs normally traverse an ATP-dependent cycle, possibly involving receptor phosphorylation, and that without ATP they accumulate in a form that cannot bind hormone. We identified such a form, the null receptor, in ATP-depleted cells. GRs are basally phosphorylated, and become hyperphosphorylated after treatment with hormone (but not RU486). In mouse receptors we have identified 7 phosphorylated sites, all in the N-terminal domain. Most are on serines and lie within a transactivation region. The time-course of hormone-induced hyperphosphorylation indicates that the primary substrates for hyperphosphorylation are the activated receptors; unliganded and hormone-liganded nonactivated receptors become hyperphosphorylated more slowly. After dissociation of substrates for hyperphosphorylation are the activated receptors; unliganded and hormone-liganded nonactivated receptors become hyperphosphorylated more slowly. After dissociation of hormone, most receptors appear to be recycled and reutilized in hyperphosphorylated form. From these and related observations, we have concluded that the postulated ATP-dependent cycle can be accounted for by hormone-induced or spontaneous dissociation of receptor-Hsp90 complexes, followed by reassociation of unliganded receptors with Hsp90 via an ATP-dependent reaction like that demonstrated in cell-free systems. Other steroid hormone receptors might traverse a similar cycle. Four of the 7 phosphorylated sites in the N-terminal domain are in consensus sequences for p34cdc2 kinases important in cell cycle regulation. This observation, along with the known cell cycle-dependence of sensitivity to glucocorticoids and other evidence, point to a role for receptor phosphorylation in controlling responses to glucocorticoids through the cell cycle.