In a study of the cues involved in spontaneous alternation in which the first turn took place in one maze and the second turn in a different maze, we found: (1) identical levels of significant, above-chance, alternation whether the mazes were parallel or perpendicular to each other; (2) blinding did not affect alternation rates; and (3) bilateral vestibular neurotomy produced a severe and long-lasting (up to 1 year) reduction of spontaneous alternation under all cue conditions tested. This reduction was greatest in the parallel condition. We conclude that, in the absence of visually differentiated maze arms, vision plays no role in two-maze spontaneous alternation while kinesthetic cues may play a role. Alternation depends on the integrity of the vestibular system, although the specificity of the vestibular alternation deficit remains to be determined. Data on righting reflexes in air, collected in evaluation of vestibular function, are also presented.