Attempts to better understand causal factors affecting estrogenicity in municipal wastewater have primarily focused on analytical evaluation of specific chemical estrogens and the use of estrogen receptor (ER) based in vitro assays. To compare analytical,in vitro, and in vivo assays for estrogenicity, wastewater from four New York and one Texas municipal wastewater facilities was evaluated for estrogenic activity using the yeast estrogen screen assay (YES) and an in vivo fish vitellogenin (VTG) assay. Estrogenic activity, as measured by the YES assay, was observed in methanol and/or methylene chloride eluents from C18 extracts in two of the New York treatment facilities and the Texas facility. Estradiol equivalents for the YES assay data ranged from </=1 to 15 ng/l. Male Japanese medaka (Oryzias latipes) were then exposed for 7 days to solvent extracts from the New York-Red Hook facility and the Texas facility. Hepatic and plasma vitellogenin were induced in medaka after exposure to the methanol eluent from the New York facility, even though the YES assay indicated that both the methanol and methylene chloride eluents were estrogenic. Whereas an estrogenic response in the YES assay was only observed in the methanol eluent from the Texas facility, plasma VTG induction was observed in both the methanol and methylene chloride eluents. In vivo estrogenic activity was nearly 10-fold greater than YES activity indicating the presence of nonestrogen receptor ligands that elicit estrogenic effects in fish through indirect mechanisms. The sole use of in vitro assays to screen for estrogenicity may underestimate estrogenic potential of wastewater.