Signal Transducer and Activator of Transcription factors (STATs) are proteins able to translocate into the nucleus, bind DNA and activate gene transcription. STATs proteins play a crucial role in cell proliferation, apoptosis and differentiation. The prevalent view is that STATs proteins are able to form dimers and bind DNA only upon phosphorylation of specific tyrosine residues in the Trans-Activation-domain. However, this paradigm has been questioned recently by the observation of dimers of unphosphorylated STATs (USTATs) by X-ray, FRET and site-directed mutagenesis. A more complex picture of the dimerization process and of the role of the dimers is, thus, emerging. Here we present an integrated modeling study of STAT3, a member of the STAT family of utmost importance in cancer development and therapy, in which we combine available experimental data with several computational methodologies such as homology modeling, protein-protein docking and molecular dynamics, to build reliable atomistic models of USTAT3 dimers. The models generated with integrative approach presented here, were, then, validated by performing computational alanine scanning for all the residues in the protein-protein interface. These results confirmed the experimental observation of the importance of some of these residues (in particular Leu78 and Asp19) in the USTAT3 dimerization process. Given the growing importance of USTAT3 dimers in several cellular pathways, our models provide an important tool to study the effects of pathological mutations at molecular/atomistic level, and in the rational design of new inhibitors of dimerization.