Numerous studies have investigated degradation of individual endocrine disrupting compounds (EDCs) in lab or natural waters. However, natural variations in water matrices and mixtures of EDCs in the environment may confound analysis of the treatment efficiency. Because chemical based analytical methods cannot represent the combined or synergistic activities between water quality parameters and/or the EDC mixtures at environmentally relevant concentrations (mu g L-1-ng L-1), bioanalytical assessments of residual estrogenic activity in treated water were used to evaluate the performance of the UV based advanced oxidation process for estrogenic contaminants in water. Four EDCs including estradiol (E-2) ethinyl estradiol (EE2), bisphenol-A (BPA) and nonylphenol (NP) were spiked individually or as a mixture at mu g L-1-ng L-1 in laboratory or natural river water. The removal rates of estrogenic activity were quantitatively evaluated by in vitro yeast estrogen screen (YES) and in vivo Vitellogenin (VTG) assays with Japanese medaka fish (Oryzias latipes). UV in combination with 10 ppm H2O2 as an oxidation process was capable of decreasing in vitro and in vivo estrogenic activity, however, in vivo estrogenic activity of the EDC mixture in natural water was not completely removed at UV fluence up to 2000 mJ cm(-2). The removal rates of in vitro estrogenic activity of the EDC mixtures were lower than those observed for single compounds, and slower in natural waters, likely due to lower steady-state concentrations of hydroxyl radicals (center dot OH) in the presence of center dot OH scavengers from the water matrix and EDC mixture. (c) 2007 Elsevier B.V. All rights reserved.