Global development of industry has been related with environmental pollution. An adverse effect on various anthropogenic substances on organisms in the environment has become a serious problem. Therefore, various methods have been developed in order to determine the toxic effect of hazardous substances on organisms and assess the risk of their presence in the environment. The risk assessment is based on the known toxicity values of individual substances that can be tested on various test organisms. Such approach has a deficiency if applied for environmental risk assessment considering that hazardous substances are not found alone in the environment but as a mixture of various substances. Substances in mixtures can act together, possibly resulting in more intense harmful effect than a cumulative effect of two pure substances. Due to the infinite number of different mixtures that can occur in the environment, mathematical models that predict their toxicity have been developed and used in risk assessment. The most commonly used model is the concentration addition model, followed by the independent action model and models resulting from their combination. The intensive development of computers with high performances supported application of advanced mathematical approaches such as quantitative structure-activity relationship (QSAR) modeling and artificial intelligence. Statistical methods used to determine the toxic effect of substances on organisms are generally divided into parametric and nonparametric methods. Whenever possible, parametric tests are used, and the most commonly ones are Student's t-test, Tukey's test, Dunnett's test and Williams' test. Every parametric test has an equivalent nonparametric test.