The extrapolation from animal data to therapeutic effects in humans, a basic pharmacological issue, is especially critical in studies aimed to estimate the protective efficacy of drugs against nerve agent poisoning. Such efficacy can only be predicted by extrapolation of data from animal studies to humans. In pretreatment therapy against nerve agents, careful dose determination is even more crucial than in antidotal therapy, since excessive doses may lead to adverse effects or performance decrements. The common method of comparing dose per body weight, still used in some studies, may lead to erroneous extrapolation. A different approach is based on the comparison of plasma concentrations at steady state required to obtain a given pharmacodynamic endpoint. In the present study, this approach was applied to predict the prophylactic efficacy of the anticholinergic drug caramiphen in combination with pyridostigmine in man based on animal data. In two species of large animals, dogs and monkeys, similar plasma concentrations of caramiphen (in the range of 60-100 ng/ml) conferred adequate protection against exposure to a lethal-dose of sarin (1.6-1.8 LD(50)). Pharmacokinetic studies at steady state were required to achieve the correlation between caramiphen plasma concentrations and therapeutic effects. Evaluation of total plasma clearance values was instrumental in establishing desirable plasma concentrations and minimizing the number of animals used in the study. Previous data in the literature for plasma levels of caramiphen that do not lead to overt side effects in humans (70-100 ng/ml) enabled extrapolation to expected human protection. The method can be applied to other drugs and other clinical situations, in which human studies are impossible due to ethical considerations. When similar dose response curves are obtained in at least two animal models, the extrapolation to expected therapeutic effects in humans might be considered more reliable.