A hypothesis is presented on the gating of ion channels. This is considered as a consequence, in part, of a large increase in viscosity of the water in the "vestibule" region of the channel in the high field present when the channel is not conducting. This part of gating amounts to "melting" of the high viscosity part of the water upon release of the field. The resulting model accounts qualitatively for a number of phenomena in the literature, including the steepness of the voltage dependence of gating, the slowing of gating upon substitution of D2O for H2O, and the pressure dependence of the gating kinetics. The viscosity increase with field is well known in the literature; several forms of electroviscous effects, a viscoelectric effect, and a generalized electrorheological effect have been described. This model appears closest to an electrorheological effect in which boundary water out to a few molecular diameters is structured in the presence of a high field, while the boundary (here, protein) moves. The size of the channel entrance is small enough for this effect to prevent conductivity. The remainder of the gating current, which occurs at more polarized potentials, is attributed to protein motion. Some consequences of the model are discussed. Qualitative comparison with published data is included.