This laboratory has been using the teleost retinal cone as a model for studying the mechanisms and regulation of retinal cell motility. In previous inhibitor studies, the authors have shown that dark-induced cone elongation requires microtubules, whereas light-induced contraction requires actin filaments. This study examines the distributions of actin filaments, microtubules, and intermediate filaments in the cone cytoskeleton. Actin filaments have been localized in isolated cones by labeling with fluorescent derivatives of phalloidin; microtubules were localized by immunofluorescent labeling with anti-tubulin. Actin, microtubule, and intermediate filament distributions have also been examined in detergent-lysed motile cell models of cones fixed with a new method that enhances preservation of the cytoskeleton. Longitudinal bundles of actin filaments extend from the cone's calycal processes through the ellipsoid and into the myoid. No actin filaments are detectable in the perinuclear region and axon, but filaments are present in both pre- and post-synaptic components of the synapse. Intermediate filaments are numerous in the perinuclear region and cone axon but relatively sparse in the myoid. In contrast, microtubule distribution is more uniform: numerous longitudinally oriented microtubules are present throughout the length of the cell. Thus the cone cytoskeleton reflects the highly polarized shape and function of the cell, with actin filaments localized to the distal movable part of the cell and intermediate filaments localized to the proximal part of the cell, which is anchored in the retina.