A study of the effects of dietary genistein on trout and sturgeon in vivo showed that sturgeon was sensitive to 20 ppm of genistein, whereas trout was not. To analyze the origin of this interspecies difference in sensitivity, a cell culture technique was developed with hepatocytes from sturgeon and compared to results obtained with hepatocytes from trout in the same system. The hepatocyte culture proved to be useful as bioassay for estrogenicity. Vitellogenin (VTG), assayed by a specific enzyme-linked immunosorbent assay, was used as a biomarker of the estrogenic activity. 17 beta-Estradiol, its glucuronide and sulfate derivatives, and estradiol analogues (ethynylestradiol and diethylstilbestrol) were tested. Nonestrogenic compounds such as androgens, progesterone, and cortisol were tested as negative controls. VTG production was monitored at doses ranging from 1 nM to 10 microM estradiol. Phytoestrogens, from the isoflavone family, were tested individually at increasing doses exhibiting dose response curves for concentrations from 500 nM to 10 microM. With tamoxifen, an antagonist of estrogen receptors, the estrogenic effect was partially reduced. The effect was the same with ICI182,780 in sturgeon, whereas the effect was the opposite in trout. The estrogenic potency of the isoflavones ranged differently between the two species in the following order: biochanin A < daidzein = formononetin < genistein < equol in trout and biochanin A < genistein < daidzein < formononetin < equol in sturgeon. Further, in sturgeon, formononetin was the most potent phytoestrogen in vitro, whereas its activity was weakest in vivo. These data suggest that one must reconsider the relevance of heterologous estrogenic tests and of homologous in vitro tests for estrogenic potency of chemicals.