The LD50 test was introduced by Trevan in 1927 for biological standardization of dangerous drugs. Since then, the LD50 has gained wide acceptance as a measure of acute toxicity of all types of substances. Recently, however, the LD50 test has been criticized as an unnecessary waste of resources. Therefore, efforts have been made to reduce the number of animals used in such tests and to avoid using this test unless required by regulations. A review of the literature has shown that a relatively small number of animals per dose level (5) and a small number of dose levels (2 or 3) are usually sufficient to calculate an LD50 and slope using moving average methods. In addition, one sex should suffice since large sex differences are seldom encountered. When a formal LD50 is not required, one of several approximate methods may be used to estimate the lethal dose. Future approaches include in vitro cytotoxicity methods and computer-based structure-activity models. The in vitro methods are still in an early stage of development and will require extensive validation before they are accepted by the toxicology community. In conclusion, when LD50 tests are required, the most economical approach should be used, without undue concern for statistical precision.