Immunotoxicology is an important aspect of the safety evaluation of drugs and chemicals. Immunosuppression, (unspecific) immunostimulation, hypersensitivity and autoimmunity are the four types of immune-mediated adverse effects. However, the nonclinical assessment of immunotoxicity is at present often restricted to animal models and assays to predict unexpected immunosuppression. There is, however, no general consensus that a variety of assays can be considered depending on the compound to be tested. A major issue is whether histological examination of the thymus, spleen, lymphoid organs and Peyer's patches is a reliable predictor of immunosuppression or whether immune function should also be assessed. A T-dependent antibody response assay, either the plaque-forming cell assay or anti-keyhole limpet haemocyanin enzyme-linked immunosorbant assay, is recommended as a first-line assay. A variety of assays, including lymphocyte subset analysis, natural killer-cell activity, lymphocyte proliferation, delayed-type hypersensitivity, cytotoxic T-lymphocyte activity and macrophage/neutrophil function assays, can also be used. In certain circumstances, host resistance assays can be considered. With the exception of contact sensitisation, very few animal models and assays can reliably predict the potential for (unspecific) immunostimulation, hypersensitivity or autoimmunity. A major limitation of immunotoxicity risk assessment is the lack of human data. Immunological end points and clinical criteria to be included in clinical trials and epidemiological studies have to be carefully standardised and validated.