The major protocol features of the immature rat uterotrophic assay have been evaluated using a range of reference chemicals. The protocol variables considered include the selection of the test species and route of chemical administration, the age of the test animals, the maintenance diet used, and the specificity of the assay for estrogens. It is concluded that three daily oral administrations of test chemicals to 21- to 22-day-old rats, followed by determination of absolute uterus weights on the fourth day, provide a sensitive and toxicologically relevant in vivo estrogenicity assay. Rats are favored over mice for reasons of toxicological practice, but the choice of test species is probably not a critical protocol variable, as evidenced by the similar sensitivity of rats and mice to the uterotrophic activity of methoxychlor. Vaginal opening is shown to be a useful, but nondefinitive, adjunct to the uterotrophic assay. The ability of test chemicals to reduce or abolish the uterotrophic response of estradiol is suggested to provide a useful extension of the uterotrophic assay for the purpose of detecting antiestrogens. The results of a series of studies on the environmental estrogen nonyl phenol (NP), and its linear isomer n-nonyl phenol, confirm that branching of the aliphatic side chain is important for activity. 17beta-Desoxyestradiol is shown to be of similar activity to estradiol in the uterotrophic assay and is suggested to represent the "parent" estrogen of NP. Benzoylation of NP and 17-desoxyestradiol did not affect their uterotrophic activity, in contrast to the enhancing effect of benzoylation on estradiol. Selected chemicals shown to be active in the immature rat uterotrophic assay were also evaluated in an in vitro yeast human estrogen receptor transactivation assay. Most of the chemicals gave similar qualitative responses to those seen in the uterotrophic assay, and the detection of the estrogen methoxychlor by the yeast assay evidenced a degree of intrinsic metabolic competence. However, the assay had a reduced ability (compared to rodents) to hydrolyze the benzoate ester of estradiol, and the estrogenic benzoate derivative of NP was not active in the yeast assay. These last results indicate that current metabolic deficiencies of in vitro estrogenicity assays will limit the value of negative data for the immediate future. The results described illustrate the intrinsic complexity of evaluating chemicals for estrogenic activities and confirm the need for rigorous attention to experimental design and criteria for assessing estrogenic activity.