Estrogen activity was measured in wastewater effluent before and after polishing via soil-aquifer treatment (SAT) using both a (hER-beta) competitive binding assay and a transcriptional activation (yeast estrogen screen, YES) assay. From the competitive binding assay, the equivalent 17alpha-ethinylestradiol (EE2) concentration in secondary effluent was 4.7 nM but decreased to 0.22 nM following SAT. The YES assay indicated that the equivalent EE2 concentration in the same effluent sample was below the method-detection limit (<2.5 x 10(-3) nM) but increased to 0.68 nM in effluent polished via SAT processes. It was hypothesized thattest-dependent differences arose because the competitive binding assay responds positively to both estrogen mimics and anti-estrogens; the YES assay responds to estrogen mimics, but test response is inhibited by anti-estrogens. The hypothesis was supported when organics extracted from wastewater effluent inhibited the YES test response to EE2 (anti-estrogenic effect). A similar extract prepared from SAT-polished effluent augmented the EE2 curve (agonist response). When hydrophobic organics in secondary effluent were fractionated, assay results indicated that several physically distinct anti-estrogens were present in the sample. From this work, it is evident that transcription-activation bioassays alone should not be relied upon to measure estrogenic activity in complex environmental samples because the simultaneous presence of both agonists and antagonist compounds can yield false negatives. Multiple in vitro bioassays, sample fractionation or tests designed to measure anti-estrogenic activity can be used to overcome this problem. It is also clear that there are circumstances under which SAT does not completely remove estrogenic activity during municipal wastewater effluent polishing.