Human aromatase (HA) is a P450 cytochrome (CYP) with an essential role in estrogen biosynthesis. Since more than 70% of breast cancers are positive for estrogenic receptor (ER), the reduction of estrogen physiological concentrations through HA inhibition is one of most important therapeutic strategies against this cancer type. Recently, experimental evidence showed that selected taxmoxifen metabolites, which are typically used as estrogen receptor modulators (SERMs), inhibit HA through an allosteric mechanism. In this work, we present a computational protocol to (i) characterize the structural framework and (ii) define the atomistic details of the determinants for the noncompetitive inhibition mechanism. Our calculations identify two putative binding sites able to efficiently bind all tamoxifen metabolites. Analysis of long-scale molecular dynamics simulations reveal that endoxifen, the most effective noncompetitive inhibitor, induces significant enzyme rigidity by binding in one of the possible peripheral sites. The consequence of this binding event is the suppression of one of the functional enzymatic collective motions associated with breathing of the substrate access channel. Moreover, an internal dynamics-based alignment of HA with six other human cytochromes shows that this collective motion is common to other members of the CYP450 protein family. On this basis, our findings may thus be of help for the development of new (pan)inhibitors for the therapeutic treatment of cancer, targeting and modulating the activity of HA and of estrogen receptor, and may also stimulate the development of new drug design strategies for chemoprevention and chemoprotection via allosteric inhibition of CYP450 proteins.