The effect of Bothrops jararacussu venom was studied in cutaneous pectoris nerve muscle preparations and in the desheathed sciatic nerve of the frog. The venom rapidly inhibited muscle twitch--tension, evoked either directly or indirectly through the motor nerve and abolished the compound action potential of the muscle and of the sciatic nerve. After fractionation of the venom by Sephadex G-50 column chromatography, all the activity was recovered in a fraction containing 30% of the total venom protein and highly enriched in two polypeptides with apparent Mr of 13-15,000, as revealed by two-dimensional polyacrylamide gel electrophoresis. The concentration of active subfraction required to obtain 50% paralysis in 1 h was 8 micrograms protein/ml. The active subfraction contained low levels of phospholipase A activity, whereas no proteolytic activity was detected. The paralyzing activity of the active subfraction on nerve-muscle preparations was not dependent on the presence of Ca2+, suggesting that phospholipase A activity is not required for the toxic effect. The active subfraction was found to cause an initial spontaneous contracture and fasciculation of the nerve-muscle preparation, and a rapid depolarization of the muscle membrane. The frequency of miniature endplate potentials was normal throughout the period of exposure to the active subfraction, although occasionally initial transient bursts were observed. At the end of the incubation, nerve endings still responded to high [K+] and to black widow spider venom. The exposure (1-2 h) to blocking concentrations of venom active subfraction provoked different degrees of morphological alteration of the muscle fibers. In contrast, no ultrastructural alterations were observed in nerve terminals, giving further support to the idea that terminals are not a prime site of the venom action. In addition to its effect on the nerve muscle-preparation, the active subfraction at higher concentrations, showed a Ca2+-dependent hemolytic activity. In the light of these results, the properties of the active subfraction of B. jararacussu venom are compared with those of other known membrane-active toxins.