To elucidate the ligand binding properties of the estrogen receptor (ER) and how ligand access to and release from the ligand binding pocket is affected by the conformational state of the receptor, we have measured the rates of estradiol association and dissociation, the equilibrium binding, and the stability of estradiol binding to denaturants, comparing wild-type human ER and a point mutant (Y537S ER) that shows full constitutive activity, i.e., the same full transcriptional activity in the absence or presence of estrogen. Ligand binding kinetics and affinity were measured with the full-length (1-595) ERs and with truncated forms of both receptors containing domains C through F (including the DNA binding, hinge, and ligand binding domains, amino acids 175-595) or domains E and F (the ligand binding domain; amino acids 304-595). With all ERs, the rates of ligand association and dissociation were considerably slower with the Y537S mutant ER than with wild-type ER (6-fold and 3-4-fold, respectively). These marked differences in ligand on and off rates for the wild-type and Y537S receptors result in a predicted (k-1/k+1) and measured Kd that is 2-fold lower for Y537S ER compared to wild-type ER. The binding of estradiol by wild-type ER was disrupted by high concentrations of urea (above 2 M), whereas the Y537S ER was distinctly more resistant to this disruption. These results are consistent with a model in which wild-type ER in the absence of ligand adopts a transcriptionally inactive collapsed pocket conformation, stabilized by specific interactions of Y537 with nearby regions of ER. When estradiol is bound, the wild-type ER adopts a transcriptionally active, closed pocket (ligand occupied) conformation. By contrast, the Y537S mutant ER favors the transcriptionally active closed pocket conformation, whether occupied by ligand or not, the latter state (closed pocket but unoccupied) accounting for its constitutive activity. Our findings suggest that the entry or exit of ligand from the binding pocket requires that ER adopt an open pocket conformation. The reduced rates of ligand association and dissociation in the constitutively active form of the ER, as well as its greater resistance to disruption of ligand binding by urea, support the supposition that the rate at which this open pocket conformation can be accessed from the unoccupied or ligand-occupied Y537S ER is slower than from the unoccupied or occupied forms of wild-type ER. Thus, the binding and release of ligand by ER require that the receptor access an open pocket state, and the ease with which this state can be accessed is affected by mutations that alter receptor conformation.