The effects of external application of micromolar concentrations of toxin 1 of the scorpion, Androctonus australis Hector, on the sodium conductance of squid giant axons have been studied quantitatively using the voltage clamp technique. Toxin concentrations which induce long plateau action potentials under current clamp conditions were found to simultaneously decrease the peak conductance and increase the delayed sodium conductance. Return to holding potential level after step depolarizations was accompanied by large exponential tails of current. The toxin-induced maintained sodium conductance increased with membrane depolarization independently of the peak conductance. Depolarizing conditioning prepulses to - 30 mV were found to almost totally inactivate the peak sodium current but to leave the delayed conductance unaffected. This property was taken as an indication that the total current is made of the added contributions of two distinct populations on sodium channels : fast activating and inactivating channels and slow activating channels. These two channel populations were separated from each other and analysed. It was found that the fast channels were almost identical to normal channels whereas the slow channels had a much slower (nearly exponential) kinetics and activated for more positive values of membrane potential. These observations strongly support the second hypothesis of Gillespie and Meves (1980) that the peak conductance and maintained conductance reflect the existence of two separate populations of channels. They further indicate that slow channels probably originate from the modification by the toxin of normal voltage-sensitive channels.