Attempts were made to determine the lethality of Formosan snake venoms in mice by various inoculation routes. No matter what inoculation routes (s.c. i.p. i.v.) and which mouse strains (ICR, NIH) were used, statistical analysis revealed that the lethality (LD50) of B. multicinctus venom was the greatest followed by N.n. atra, Tr. gramineus, Tr. mucrosquamatus and Agk. acutus venom. The i.v. inoculation resulted in the highest lethality of haemorrhagic venoms (Tr. gramineus, Tr. mucrosquamatus, Agk. acutus) and the i.p. was the next, while in the case of neurotropic venoms (B. multicinctus, N.n. atra) the results were reversed, moreover, the s.c. was the lowest. The reproducibility of different inoculation routes on determining the lethality of venom was expressed by coefficient of variation (S.D./mean). The i.p. was the best inoculation route for haemorrhagic and N.n. atra venoms and the i.v. was the best for B. multicinctus venom. The results obtained from mouse ICR strain seemed to be more reproducible than that obtained from NIH strain. The best method for determining the neutralizing capacity of B. multicinctus antivenin was the i.p. inoculation in ICR mouse.