Several methods for treatment of data from toxicity tests to determine the median lethal concentration (LC50) are discussed. The probit and logit models widely used for these calculations have deficiencies; therefore, a calculational method, named the 'trimmed Spearman-Karber method', is developed. Examples of actual and hypothetical bioassay test data are given, and comparisons are made of the abilities of these three methods to treat these data. The trimmed Spearman-Karber method is not subject to the problems of the probit and logit models, has good statistical properties, is easy to use, and is recommended for accurate and precise calculation of LC50 values and their 95% confidence interval end points.