Positively charged voltage sensors of sodium and potassium channels are driven outward through the membrane's electric field upon depolarization. This movement is coupled to channel opening. A recent model based on studies of the KvAP channel proposes that the positively charged voltage sensor, christened the “voltage-sensor paddle”, is a peripheral domain that shuttles its charged cargo through membrane lipid like a hydrophobic cation. We tested this idea by attaching charged adducts to cysteines introduced into the putative voltage-sensor paddle of Shaker potassium channels and measuring fractional changes in the total gating charge from gating currents. The only residues capable of translocating attached charges through the membrane-electric field are those that serve this function in the native channel. This remarkable specificity indicates that charge movement involves highly specialized interactions between the voltage sensor and other regions of the protein, a mechanism inconsistent with the paddle model.