Metastatic cancer of unknown primary site (CUP syndrome) comprises 2-5% of all solid malignant tumors. One should distinguish between initial CUP (primary tumor later detected) and the true CUP syndrome (primary tumor remains unknown for a patient's lifetime despite thorough diagnostic work-up). For initial CUP, the most important auxiliary diagnostic method is immunohistochemistry, which should be applied in a two-step algorithmic fashion. Firstly, a small marker panel (including certain cytokeratins) yields a preliminary categorization of the tumor. Secondly, selective, organ-specific markers (including recently established markers such as TTF-1 and uroplakin) and further tumor group markers may further subclassify or even identify the primary tumor. Although they are a heterogeneous group, true CUP tumors share some unique biological features such as an early metastatic phenotype and unusual metastasis patterns, and they mostly have a very poor prognosis. Even autopsy reveals the primary site in only 55-80% of cases, most commonly in the lung and pancreas. True CUP tumors, predominantly adenocarcinomas and poorly differentiated carcinomas, may exhibit unusual immunohistochemical phenotypes. Nevertheless, careful histologic and immunohistochemical examination are essential not only for determining the actual tumor immunophenotype but in particular for identifying therapy-responsive subgroups such as neck lymph node CUP, axillary lymph node CUP of females, neuroendocrine CUP, and germ cell tumor CUP of males. For CUP syndrome, future interdisciplinary research efforts are needed, such as gene expression profiling using microarrays. It is thus to be hoped that pathology will contribute to the elucidation of the largely still enigmatic pathogenesis of the CUP syndrome, to improve its diagnosis and classification and, finally, to aid in the development of more specific therapeutic regimens.