Most in vitro reporter gene assays used to assess estrogenic contamination are based on human estrogen receptor α (hERα) activation. However, fish bioassays can have distinct response to estrogenic chemicals and mixtures, questioning the relevance of human-based bioassays for assessing risk to this species. In this study, zebrafish liver cells stably expressing zebrafish ERβ2 (ZELHβ2) and human breast cancer cells expressing hERα (MELN) were used to quantify the estrogenic activity of 25 surface water samples of the Danube River, for which chemicals have been previously quantified. Most samples had a low estrogenic activity below 0.1 ng/L 17β-estradiol-equivalents that was more often detected by MELN cells, while ZELHβ2 response tend to be lower than predicted based on the chemicals identified. Nevertheless, both bioassays quantified well a higher estrogenic activity at two sites, which was confirmed in vivo using a transgenic zebrafish assay. The results are discussed considering the effect-based trigger values proposed for water quality monitoring.