The accuracy of the docking of a ligand in its modeled binding-site depends on the reliability of this model. To provide a model with experimental support, we have developed an engineered affinity labeling method combining cysteine-reactive probes with substitutions of putative site-lining residues into cysteines. This strategy amounts to building chemical sensors for the proximity of the substituted cysteines; it requires an activity or binding assay to monitor the irreversible occupancy of the site by the reactive ligand. Using affinity probes made reactive in different positions, the docking of the ligand can be inferred from the observed pattern of coupling reactions. The method involves three steps: ligand chemistry, mutagenesis and biological assays, which are detailed and scrutinized in the review: lead selection, ligand derivatization, and evaluation of the affinity probes (stability, reactivity and biological properties) for the ligands; positional selection and mutant properties for the cysteine substitutions; functional controls and assays for the analysis of the irreversible reactions. Examples illustrate the different criteria of concern; the data are interpreted in terms of binding-site structure and function. Potentially, the method can explore protein dynamics, since its targets are full-length membrane-inserted heteromeric proteins: it can detect subtype-dependent or activation-induced conformational changes.