Laboratory data from Milan and Houston were evaluated to determine the extent to which the distribution of estrogen receptor (ER) and progesterone receptor (PgR) has changed with time. Results from over 11,500 ER and over 8,200 PgR determinations (6,194 ER and 3,127 PgR from Milan) were analyzed. All assays in Milan were performed by a dextran-coated charcoal method and in Houston by a sucrose density-gradient method. The data demonstrate a time-dependent, upward drift in the amount of ER and PgR detected, with the effect most pronounced at the lower end of the distribution curves. We attribute this change to optimization of all facets of the receptor assay procedures (tissue harvesting and storage) as well as to a change in breast cancer biology. These results suggest that studies correlating certain biological parameters with receptor status (whether using qualitative or quantitative scales) need to be re-examined. For example, a population of 349 node-negative patients who did not receive any adjuvant treatment was studied in Milan to determine any association between disease-free survival (DFS) and receptor status. If the "historical" threshold values (10 fmol/mg protein) were used to determine receptor status, no significant difference in DFS at 5 years was detected. Even the combination of ER and PgR did not improve the predictive power of receptor status. In the premenopausal subgroup, ER status did predict the 5-year DFS. However, if the threshold value for PgR was adjusted to 25 fmol/mg protein, patients with ER-positive, PgR-positive tumors had significantly better 5-year DFS than patients with ER-negative, PgR-negative tumors. In addition, PgR status alone was associated with significantly improved 3-year DFS if the subgroups of PgR less than 5 fmol/mg protein and PgR greater than 100 fmol/mg protein were compared. We conclude from these data that: 1) historical threshold values for receptor positivity should be re-examined in all laboratories; 2) studies involving receptor results determined over an extended period of time should attempt to "normalize" these results; and 3) the quantitative assessment of receptor status should be used whenever possible.