The LD50-test was developed in 1927 for the biological standardization of dangerous drugs. Then it was incorporated into the routine toxicological protocol of other classes of chemical compounds and is now part of practically all governmental guidelines which regulate toxicological testing of chemicals. For scientific, economic, and ethical reasons it is necessary to periodically reassess all toxicological test procedures, including the LD50-test. Tests which are not optimal or that have become obsolete because of new scientific knowledge, must be changed or eliminated. The review of the LD50-test shows that the precision of the procedure is dependent on the number of animals used. But even with large numbers of animals there are considerable variations of the test results, because the numerical value of the LD50 is influenced by many factors, such as animal species and strain, age and sex, diet, food deprivation prior to dosing, temperature, caging, season, experimental procedures, etc. Thus, the LD50 value cannot be regarded as a biological constant. Through standardization of the test animals and the experimental conditions the variability of the LD50 determinations can be reduced but never fully eliminated. There are several tests with which an approximate LD50 can be determined. These methods use fewer animals than the classical LD50-test, but their precision and reproducibility are sufficient for most purposes of acute toxicity testing. Through incorporation of physiological, hematological, biochemical, pathological, and histopathological investigations in the simplified test procedures with small numbers of animals, it is possible to markedly increase the informational content of the results with regard to the toxicological spectrum and the target organs of toxicity. Such studies have already replaced the LD50-test in large animals, such as dogs and monkeys. It is also desirable to replace the LD50 in rodents with such a procedure. With pharmacologically inert compounds that have no acute effects with single administration the classical LD50-test does not provide relevant toxicological results. For the prediction of the human lethal dose and for the prediction of the symptomatology of poisoning after acute overdosing in man the LD50-test is of limited usefulness. An acute toxicity test with small numbers of animals combined with comprehensive studies of physiological functions, biochemical and histopathological examinations often provides more important information for emergency physicians and poison control centers. For the selection of doses to be used in subacute and chronic toxicity experiments the LD50-test does not provide consistent and reliable results. A simple pilot experiment with few animals but repeated dosing gives more useful information. For the evaluation of special risks for the human newborn and infant the LD50-test is poorly suited. For the appraisal of pharmacokinetic behavior and bioavailability the LD50-test gives only semi-quantitative, often ambiguous information. In all cases where the acute toxicity testing is mainly concerned with the evaluation of toxicological potential of the test substances, the symptomatology following acute overdosing, and the knowledge of target organs of toxicity, the classical LD50-test should be replaced by a more comprehensive short term test that can be done with small numbers of animals. The classical LD50-test should only be permitted in those rare instances where a high precision of the LD50 determination is indispensable.