Multilayered, lipid-rich myelin increases nerve impulse conduction velocity, contributes to compact nervous systems, and reduces metabolic costs of neural activity. Based on the hypothesis that increased impulse conduction velocity provides a selective advantage that drives the evolution of myelin, we simulated a sequence of plausible intermediate stages of myelin evolution, each of which providing an enhancement of conduction speed. We started with the expansion of insulating glial coverage, which led first to a single layer of myelin surrounding the axon and then to multiple myelin wraps with well-organized nodes. The myelinated fiber was modeled at three levels of complexity as the hypothesized evolutionary progression became more quantitatively exacting: 1) representing the fiber as a mathematically-tractable uniform active cylinder with the effect of myelination approximated by changing its specific capacitance (C(m)); 2) representing it as a chain of simple, cable-model compartments having alternating nodal and internodal parameters subject to optimization, and 3) representing it in a double cable model with the axon and myelin sheath treated separately. Conduction velocity was optimized at each stage. To maintain optimal conduction velocities, increased myelin coverage of axonal surface must be accompanied by an increase in channel density at the evolving nodes, but along with increases in myelin thickness, a reduction in overall average channel density must occur. Leakage under the myelin sheath becomes more of a problem with smaller fiber diameters, which may help explain the tendency for myelin to occur preferentially in larger nerve fibers in both vertebrates and invertebrates.