Size-specific interaction of alkali metal ions with aromatic side chains has been proposed as a mechanism for selectivity in some K+ channel proteins. Experiments on gas-phase cluster ions of the form M+(C6H6)(n) (H2O)(m), with M = Na or K, have demonstrated that the interaction between benzene and K+ is sufficiently strong to result in partial dehydration of the ion, i.e., benzene will displace some water molecules from direct contact with the ion. In sharp contrast, there is no evidence that benzene can displace water from the first hydration shell of Na+. The resistance of Na+ (H2O)(4) towards dehydration in an aromatic environment suggests a molecular-level mechanism for the low permeability of Na+ through the pore region of K+ channel proteins: the hydrated Na+ ion is too large to pass, while K+ can shed enough of its hydration shell to fit through the pore. These results also suggest that it may be possible to design a new class of ionophores that take advantage of the cation-pi interaction to confer ion selectivity. This is the first experimental evidence that K+ selectively interacts with an aromatic complex in an aqueous environment, while Na+ does not. A remarkable sidelight from this study was the discovery of a self-assembled cluster ion, Na+ (C6H6)(8)(H2O)(4), with a single structure: an inner shell of four water molecules and an outer layer of eight benzene molecules, each of the latter fixed by a pi-hydrogen bond to one of the eight interior O-H groups.