Experiments on sodium channel inactivation kinetics were performed on voltage-clamped crayfish giant axons. The primary goal was to investigate whether channels must open before inactivating. Voltage-clamp artifacts were minimized by the use of low-sodium solutions and full series resistance compensation, and the spatial uniformity of the currents was checked with a closely spaced pair of electrodes used to measure local current densities. For membrane potentials between -40 and +40 mV, sodium currents decay to zero with a single exponential time-course. The time constant for decay is a steep function of membrane potential. The time-course of inactivation measured with the double-pulse method is very similar to the decay of current at the same potential. Steady-state inactivation curves measured with different test pulses are identical. The time-course of double pulse inactivation shows a lag that roughly correlates with the opening of sodium channels, but detailed comparisons with the time course of the prepulse current suggest that it is not strictly necessary for channels to open before inactivating. Measurements of the potential dependence of the integral of sodium conductance area also inconsistent with the simplest cases of models in which channels must open before inactivating.