The admittance of the squid giant axon membrane has been measured, using an intracellular electrode, at frequencies up to 40 MHz. The existence of a radio frequency dispersion, previously detected with extracellular electrodes (Cole, 1976) and attributed to the Schwann cell layer, has been confirmed and followed to higher frequencies. For a comparable method of analysis, membrane parameters similar to those given by Cole (1976) have been calculated. The radio frequency dispersion has a centre frequency at approximately 1.8 MHz, and the properties of a parallel combination of a 28 nF cm-2 capacity and a 3.3 omega cm2 resistance. When the axon membrane capacity is calculated, taking into account the radio frequency dispersion, as described above, the capacity remains frequency dependent throughout the range studied. If it is assumed that at high frequencies the axolemma capacity becomes constant at approximately the value for a lipid bilayer, a radio frequency dispersion is found which cannot be accounted for in terms of a simple equivalent circuit with two passive components, but appears to arise from a network with a distribution of relaxation times. This result could be consistent with the morphology of the Schwann cell layer. The radio frequency dispersion referred to in (4) can be described reasonably well by a circuit with two dispersions having centre frequencies of 250 kHz and 3.2 MHz respectively. The corresponding axolemma capacity (100-500 kHz) would be approximately 0.6 microF cm-2. It is argued that between 50 and 100 kHz the geometrical capacity arising from the non-polar regions of the membrane is a major contributor to the axon membrane capacity, and that capacity variations arising from compositional changes in the lipid bilayer are best monitored in this frequency range.