Classification, pathogenesis and molecular pathology of primary central nervous system lymphomas (PCNSL) pose major clinico-pathologic problems. Application of histopathologic classification schemes developed for nodal lymphomas, i.e. the Kiel Classification and the Working Formulation, is not reliable and clinically not relevant for PCNSL. The more recent REAL Classification will simplify subtyping of PCNSL, but reliability and clinical significance remain to be determined. There are virtually no experimental data on whether PCNSL develop outside of the brain, or whether they arise from polyclonal lymphoproliferations within the brain. Mutations of oncogenes and tumor suppressor genes have been analyzed in only a few tumors, and their type and frequency is essentially unknown. In conclusion, compared with both neuroectodermal brain tumors and nodal lymphomas, and given the increasing incidence and clinical impact of PCNSL, there is a remarkable lack of histopathologic consensus as well as a surprising shortage of pathogenetic and molecular genetic information.