Ultraviolet irradiation of the estrogen binding protein in rat uterine cytosol results in a progressive photoinactivation which is rapid at 254 nm and slower at greater than 315 nm. Both unfilled and estradiol-filled sites are inactivated at approimately the same rates at 254 nm (t 1/2 equals 8 min and 11 min, respectively), but at 315 nm, empty sites are consumed much more rapidly (t 1/2 equals 3.4 hr) than filled ones (t 1/2 equals 24 hr). The protective effect of the estrogen ligand at this wavelength appears to depend on its binding to the estrogen-specific binding site, as inactivation rate studies at different concentrations of estrone, estradiol, and estriol show a good correlation between the extent of protection and the fractional saturation of the high affinity estrogen binding sites. Scatchard analysis indicates that inactivation is the result of a loss of binding sites and not a decrease in their affinity, and sedimentation analysis shows that increased heterogeneity and aggregation of the estrogen binding species accompanies the photoinactivation process. Photoinactivation appears to be the result of direct irradiative damage of the animo acid residues, as the inactivation rate is the same under air and nitrogen atmospheres, and is unaffected by nucleophiles, reductants, and radical scavengers. When photoinactivation is measureed by irradiation of cytosol containing [3-H]estradiol, a concurrent photocovalent attachment process is noted; the steroid becomes linked to protein in a solvent-extractable manner (boiling ethanol inextractable). This attachment, however, does not appear to be related to the steroid binding at the estrogen binding site. Its rate is affected by reductants and scavengers. A similar photocovalent attachment reaction occurs when bovine serum albumin or ovalbumin is irradiated in the presence of [3-H]estradiol or [3-H]diethylstilbestrol. The detailed reactions involved in this photocovalent attachment process have not been defined at present.