OBJECTIVE In order to evaluate the estrogenic activity of sediments and XAD water extracts of selected sites of the catchment area of the River Neckar, a river system in Southern Germany, an integrative assessment approach was used to assess the ecological hazard potential of endocrine-disrupting compounds in sediment and water. METHODS The approach is based on estrogen receptor-mediated vitellogenin synthesis induced in isolated hepatocytes of rainbow trout and quantified in a non-radioactive dot blot/RNAse protection-assay in parallel to comprehensive chemical analyses of estrogenic substances. CONCLUSIONS Numerous investigated extracts revealed an estrogen activity comparable to that of the positive control (1 nM 17beta-estradiol corresponding to 270 ng/L in the test medium). Based on a concentration factor of 30 in the extracts and a recovery of XAD resins of approximately 80%, 17beta-estradiol equivalent concentrations between 20 and 26.7 ng/L could be calculated downstream of a sewage treatment plant (< 0.1 ng/L for a reference site). A comparison of the bioassay-derived Bio-TEQs (toxicity equivalents) and the Chem-TEQs revealed a high correlation with a Pearson coefficient of 0.85, indicating that the same ranking of the samples could be obtained with respect to the endocrine disrupting potential with both chemical and bioanalytical analysis. However, the TEQ concentrations computed from chemical analyses were significantly lower than the bioassay-derived TEQ concentrations. In fact, in none of the samples, more than 14% of the vitellogenin-inducing potency could be attributed to the substances (steroids, alkylphenols, bisphenol A, diethylstilbestrol) analyzed. A comparison of the endocrine disrupting potential of sediments extracted by the solvents acetone and methanol revealed lower biological effects for acetone-extracted samples. Possible reasons may be a masking of endocrine effects in acetone extracts by cytotoxicity, a low extraction efficiency of the solvent acetone, or anti- estrogen potencies of some extracted sediment compounds. Using a mass balance approach, the contribution of the compounds analyzed chemically (Chem-TEQs) to the total endocrine activity (Bio-TEQs) was calculated. Based on the very low detection limits, particularly of the steroids with their high TEF factors, results revealed that a calculation of the Chem-TEQs is associated with considerable scale inaccuracy: Whereas only 7-15% of the biological effectiveness (Bio-TEQs) could be explained by endocrine substances identified above the detection limits, the assumption of concentrations slightly below the given detection limits would result in a significant over estimation (137-197%) of the Bio-TEQs. Even the interassay variation of the dot blot assay with different fish donors for primary hepatocyte (factor 2-2.5) is relatively low, when compared to the large range of the Chem-TEQ concentrations (factor 20) obtained when applying different modes of calculation. CONCLUSIONS Overall, only a minor portion of the endocrine activity detected by bioassays could be linked to compounds identified by chemical analysis. In vitro assays for assessment of endocrine activities are useful as sensitive integrating methods that provide quantitative estimates of the total activity of particular receptor-mediated responses. Although discrepancies may also result from different bioanalytical approaches, it is overall likely that bioanalytical and not chemical analytical approaches give the correct estimate of endocrine disrupting potencies in environmental samples. As a conclusion, assessment of endocrine disruption based on chemical analysis alone does not appear sufficient and further research into the spectrum of substances with potential endocrine activity as well as into additive or even synergistic effects in complex environmental samples is urgently needed.