Permeabilities of squid axon membranes to various cations at rest and during activity have been measured by voltage clamp before and during internal perfusion of 4 X 10(-5) M grayanotoxin I. The resting sodium and potassium permeabilities were estimated to be 6.85 X 10(-8) cm/sec and 2.84 X 10(-6) cm/sec, respectively. Grayanotoxin I increased the resting sodium permeability to 7.38 10(-7) cm/sec representing an 11-fold increase. The potassium permeability was increased only by a factor of 1.24. The resting permeability ratios as estimated by the voltage clamp method before application of grayanotoxin I were Na (1): Li (0.83): formamidine (1.34): guanidine (1.49): Cs (0.87): methylguanidine (0.86): methylamine (0.78). Grayanotoxin I did not drastically the resting permeability ratios with a result of Na (1): Li (0.95): formamidine (1.27): guanidine (1.16): Cs (0.47): methylguanidine (0.72): methylamine (0.46). The membrane potential method gave essentially the same resting permability ratios before and during application of grayanotoxin I if corrections were made for permeability to choline as the cation substitute and for changes in potassium permeability caused by test cations. The permeability ration choline/Na was estimated to be 0.72 by the voltage clamp method and 0.65 by the membrane potential method. Grayanotoxin I decreased the ration to 0.43. The permeability ratios during peak transient current were estimated to be Na (1): Li (1.12): formamidine (0.20): guanidine (0.20): Cs (0.085): methylguanidine (0.061): methylamine (0.036). Thus the sodium channels for the peak current are much more selective to cation than the resting sodium channels. It appears that the resting sodium channels in normal and grayanotoixn I-treated axons are operationally different from the sodium channels that undergo a conductance increase upon stimulation.