Test systems to screen for estrogenicity and appropriate biomarkers of human exposure are required for epidemiological studies of endocrine disruption. We addressed these issues by developing and standardising a method to assess the total estrogenic xenobiotic burden in human adipose tissue. In this study, which is the continuation of a previous work, we have improved the protocol for extensive fractionation of a higher number of tissue samples in order to investigate bioaccumulated xenoestrogens that are candidates for estrogenicity and to assess their combined estrogenic effect. This was achieved by extensive HPLC separation of xenoestrogens from endogenous hormones followed by testing of individual fractions in the E-Screen test for estrogenicity. Organochlorine pesticides, PCBs and halogenated bisphenols and alkylphenols were collected in the most lipophilic fractions, followed by progestins, androgens and estradiol esters, and then by steroidal estrogens; phyto- and myco-estrogens were collected around the end of the run. These results were confirmed by exhaustive chemical analysis. In 458 human adipose tissue samples, the total effective xenoestrogen burden was positive in 75% of samples in the pooled fraction that contained organohalogenated xenoestrogens (mean 515.3 pM Eeq/g lipid; range 0-14.5 nM) and in 82% of samples in the pooled fraction where natural estrogens eluted (mean 696.6 pM Eeq/g lipid; range 0-12.9 nM). Organochlorine pesticides emerged as candidate chemicals for the estrogenicity of the first pooled fraction, because DDT and derivatives were present in 98.3% of the samples. However, no correlation was found between the concentration of any single chemical and the estrogenicity determined in the bioassay. There may be several reasons for this lack of concordance: (i) the estrogenic effects depicted in the E-Screen bioassay are a consequence of the combined effect of several organohalogens or (ii) the proliferative effect is due to other chemicals not measured. Because additive, synergistic or antagonistic mechanisms may account for the final effect observed in the pooled fractions, the approach proposed in this work is more appropriate for exposure assessment in epidemiological studies than the determination of individual chemicals in human samples.