The pores of ion channel proteins are often modeled as static structures. In this view, selectivity reflects rigidly constrained backbone orientations. Such a picture is at variance with the generalization that biological proteins are flexible, capable of major internal motions on biologically relevant time scales. We tested for motions in the sodium channel pore by systematically introducing pairs of cysteine residues throughout the pore-lining segments. Two distinct pairs of residues spontaneously formed disulfide bonds bridging domains I and II. Nine other permutations, involving all four domains, were capable of disulfide bonding in the presence of a redox catalyst. The results are inconsistent with a single fixed backbone structure for the pore; instead, the segments that line the permeation pathway appear capable of sizable motions.