The Herzfeld criterion for an insulator-to-metal transition is recast here to predict the critical concentration (n(cr)) for metallization of a solution having a solute more polarizable than the host solvent. The Clausius-Mossotti relation applied to the pure solvent shows that its polarizability and density affect n(cr) only through the pure solvent refractive index or high-frequency dielectric constant. The critical concentration is reduced by solvent polarizability but increased by positive solute partial molar volumes. The electric dipole polarizabilities of the alkali metal anions Li-, Na-, and K- in solution are estimated by comparing with the polarizability changes induced on dissolving other ions. The n(cr) predicted for numerous nonaqueous solutions containing Li-, Na-, and K-coupled with the usual Herzfeld criterion explains why condensed phases containing either Na- and K- and cations complexed by crown ethers or cryptand 222 are usually nonmetallic. The critical composition (gamma(cr)) for metallization of the liquids (Li+(CH3NH2)(y))Na- is predicted to be 5.3 agreeing with the experimental value between 5 and 6. For the condensed phases (Li+(CH3CH2NH2)(y))Na-, a y(cr) of around 3 is predicted, explaining their nonmetallic nature for y = 4 and suggesting that the liquids with compositions close to y = 3 should be reexamined experimentally. Although the polarizability of Na- in the insulating solid (Li+(NH2CH2CH2NH2)(2))Na- is not known, the y(cr) of 2.35 resulting from using the overestimated anion polarizability taken from (Na+C222)Na- suggests the possibility of an insulator-to-metal transition induced by applied pressure.