We compared three radioligands for use in a cytoplasmic estrogen-receptor assay, using pooled cytosol from human breast adenocarcinomas. The estrogen receptor content was determined in vitro by a dextran-coated charcoal method involving a 4-h incubation with and without diethylstilbestrol. Tritiated moxestrol failed to come to equilibrium in 4 h, thereby preventing the use of conventional one-component Scatchard analysis. The use of 125I-labeled estradiol resulted in a higher estimate of estrogen-receptor concentration than that obtained with use of tritiated estradiol. This overestimation was not corrected by Scatchard, double-reciprocal, or Woolf plots, or by two different methods of data analysis: least squared and "robust." An underestimation of the specific activity of iodoestradiol with respect to that of tritiated estradiol and an unrecognized second component of nonreceptor binding could explain this disparity.