Glucocorticoids induce tyrosine aminotransferase (TAT) in hepatoma cells. We have previously shown that both the concentration of the agonist dexamethasone (Dex) required for half-maximal induction (EC50) and the amount of agonist activity produced by the antagonist dexamethasone 21-mesylate (Dex-Mes), expressed as a percentage of maximum induction achieved by Dex, are different in Fu5-5 and HTC cells. Furthermore, both activities vary over several weeks in each cell line in an apparently random manner, but, nevertheless, are correlated by a linear semilog plot. We now find that this long term and previously unpredictable variation in both the Dex EC50 and the amount of Dex-Mes agonist activity for the induction of TAT enzyme activity can be made to occur reproducibly in 40 h or less by changing the cell density and/or amount of medium in the tissue culture plates. Thus, a higher cell density and/or a lower volume of medium produced both higher amounts of Dex-Mes agonist activity and lower EC50 values for Dex. Experiments with cells at different densities but exposed to the same medium indicated that cell density was the dominant determinant. A qualitatively identical modulation was seen at the level of TAT mRNA, but not mouse mammary tumor virus RNA. We are not aware of any previous report of cell growth conditions altering either the level of agonist activity of an antisteroid or the EC50 of a full agonist. These results further indicate that extrachromosomal parameters, such as cell-cell contact and/or a diffusable factor(s), can modulate the basic features of glucocorticoid induction of some, but not all, glucocorticoid-inducible genes.