Since the 1990s there has been an ever-increasing scientific and societal concern about the possible adverse effects caused by endocrine disruptors (ED). In this context, the estrogen receptor (ER) represents a well-studied biological target since its biological function can be disrupted by several chemical families. In vivo testing techniques can be used to address fundamental questions on the adverse effects of chemicals that target this receptor. Nevertheless, a systematic use of these experimental approaches would lead to high monetary and ethical costs. Indeed, the number of chemicals that can be found in the environment is particularly large. Therefore, the coupling between in silico and in vitro tools represents an interesting option since it can provide an insight into fundamental aspects of endocrine disruption without the need of conducting experiments on animals. Within this strategy, the computational models prioritize chemicals on the basis of their binding affinities for the ER in order to optimize the subsequent use of in vitro methods. We analyzed the pertinence of such a testing strategy by means of in silico predictions yielded by freely available computational tools (e.g. Endocrine Disruptome, OECD QSAR Toolbox, QSARINS) and by means of original and bibliographic experimental data from in vitro assays based on the stable expression of a bioluminescent reporter gene (luciferase) under the transcriptional control of the ER. A preliminary comparison between in silico and in vitro results showed that a preparatory computational filter allows the final shortlist of chemicals to be enriched in true positives. Further study is ongoing to clarify the optimal criteria and tools to be adopted for an effective in silico screening.