The cross-reactivity of IgE antibodies is of interest for various reasons, three of which are discussed. Firstly, from the clinical view, it is important to know the patterns of cross-reactivity, because they often (but not always) reflect the pattern of clinical sensitivities. We discuss the cross-reactivities associated with sensitization to pollen and vegetable foods: PR-10 (Bet v 1-related), profilin, the cross-reactive carbohydrate determinant (CCD), the recently described isoflavone reductase, and the (still elusive) mugwort allergen that is associated with celery anaphylaxis; cross-reactivities between allergens from invertebrates, particularly tropomyosin, paramyosin, and glutathione S-transferase (GST); and latex-associated cross-reactivities. Clustering cross-reactive allergens may simplify diagnostic procedures and therapeutic regimens. Secondly, IgE crossreactivity is of interest for its immunologic basis, particularly in relation to the regulation of allergic sensitization: are IgE antibodies to allergens more often cross-reactive than IgG antibodies to ''normal'' antigens? If so, why? For this discussion, it is relevant to compare not only the structural relation between the two allergens in question, but also the relatedness to the human equivalent (if any) and how the latter influences the immune repertoire. Thirdly, prediction of IgE cross-reactivity is of interest in relation to allergic reactivity to novel foods. Cross-reactivity is a property defined by individual antibodies to individual allergens. Quantitative information (including relative affinity) is required on cross-reactivity in the allergic population and with specific allergens (rather than with whole extracts). Such information is still scarce, but with the increasing availability of purified (usually recombinant) allergens, such quantitative information will soon start to accumulate. It is expected that similarity in short stretches of the linear amino-acid sequence is unlikely to result in relevant crossreactivity between two proteins unless there is similarity in the protein fold.