The background of possible selectivity–affinity correlations and their limitations is reviewed, with typical crown ether and cryptand complexes, ionic associations, hydrogen bonded complexes and complexes driven by van der Waals, stacking or hydrophobic interactions, with some additional topics including associations based on metal coordination as supplementary material. This tutorial review is addressed to students and researchers interested in molecular recognition, and relates to the design of sensors, of discriminators for separation processes, of supramolecular devices and of drug compounds. A theoretical analysis of selectivity in supramolecular host–guest complexes, defined as a difference in binding free energies for structurally related guests, as a function of total binding free energy shows that for certain types of intermolecular interactions one may observe a correlation between selectivity and affinity. Such correlation fails however if the selectivity is due to additional interactions at a secondary binding sites, which is expected in complexes with anisotropic guest molecules. Several clear examples of theoretically expected selectivity–affinity correlations are found. The influence of reaction conditions on the experimentally observed selectivity, defined as a difference in complexation degrees with different guests in the presence of added receptor, is illustrated. The importance of often neglected solvent effects on selectivity is exemplified with ionophore and hydrogen bonded complexes.

Addition of excess of NN-dimethylbenzylamine to a chloroform solution of [Pd₃(O₂CMe)₆] leads to instantaneous depolymerisation of the trimer giving the monomer trans-[Pd(O₂CMe)₂(PhCH₂NMe₂)₂]. Reversible dissociation of the amine from the latter affords a pseudo-three-co-ordinate 14-electron intermediate [Pd(O₂CMe)₂(PhCH₂NMe₂)] which undergoes subsequent rate-limiting ortho-palladation to form the cyclopalladated acetate-bridged dimer [{Pd(O₂CMe)(C₆H₄CH₂NMe₂)}₂]. The rate-limiting step is electrophilic in character; the slope of the corresponding Hammett plot for differently ring-substituted NN-dimethylbenzylamines is –1.6. The kinetic isotope effect, k_H/k_D, for PhCH₂NMe₂ is 2.2 ± 0.2. The activation entropy for the rate-limiting step is very negative, ca. –250 J K^–1 mol^–1 for PhCH₂NMe₂, suggesting a highly ordered tight transition state. This dissociative path is a factor of ca. 100 faster than a parallel one, involving the 16-electron monomer trans-[Pd(O₂CMe)₂(PhCH₂NMe₂)₂] but without loss of the amine. Ortho-palladation is not rate-limiting in acetic acid solvent, where slow cleavage of acetate bridges in polynuclear palladium species occurs, affording a vacant co-ordination site for subsequent rapid ortho-palladation. A comparison of intra- and inter-molecular activations of carbon-hydrogen bonds in arenes by palladium(II) in terms of ‘effective molarities’ shows that the ratio k_intra/k_inter is not less than 3.6 × 10² mol dm^–3.