In silico or computational tools could be used more effectively in endocrine disruptor risk assessment for prescreening potential endocrine disruptors, improving experimental in vitro screening assay design and facilitating more thorough data analyses. The in silico tools reviewed here are three-fold and include the use of: (1) nuclear receptor (NR) crystal structures and homology models to examine potential modes of ligand binding by different representative compounds; (2) multivariate principal component analyses (PCA) techniques to select best predicted cell lines for endocrine disrupting chemicals (EDC) risk assessment purposes; (3) NR quantitative structure-activity relationships (QSARs) that can be constructed from varied biological data sources, using multivariate partial least squares (PLS) techniques and specific descriptors. The cytosolic and NR examples discussed here include the Ah receptor, (AhR), the human oestrogen receptor alpha (hERalpha) and the human pregnane X receptor (PXR). The varied biological data sets can be compared to give a more integrated dimension to receptor cross talk mechanisms, with further support from molecular modelling studies.