ERV~ cells display a number of unusual properties which are generally viewed as adaptations to their extreme length and regional specialization. The dense and highly organized neurofilament (NF) ~ and microtubule (MT) arrays which serve to support nerve axons provide one example; another is the phenomenon of axonal transport, by which macromolecules synthesized in the cell body move outward along the axon in several relatively discrete components with widely different velocities (3, 22, 32, 95; Fig. 1). During the 1970s, a crucial connection was made between these two properties of nerve cells by the demonstration that the slowest moving component of axonal transport (called slow component a [SCa]) conveys primarily tubulin and the three proteins comprising NFs (18, 23, 27, 33, 41, 52, 78, 96). This was achieved by a synthesis of the biochemical and ultrastructural data available at that time; since then, intensive study of neuronal cytoskeletal proteins has generated a succession of models of their transport, assembly, and turnover (39, 43, 58, 83, 93, 94). Recently, a number of studies have indicated that the components and interactions of the neuronal cytoskeleton may be more complex than previously thought. This review will attempt to evaluate whether current models are sufficient to accommodate the results of radiolabeling studies of axonal transport, ultrastructural studies, and recent work on the dynamics of the MT and NF systems. The principal questions to be addressed are (a) in what state are cytoskeletal proteins transported in axons; (b) how are their assembly and interactions regulated; and (c) in what region or regions of the axon does assembly of the cytoskeleton occur?