The hypothesis that viruses or other infectious agents may cause schizophrenia or bipolar disorder dates to the 19th century but has recently been revived. It could explain many clinical, genetic, and epidemiologic aspects of these diseases, including the winter-spring birth seasonality, regional differences, urban birth, household crowding, having an older sibling, and prenatal exposure to influenza as risk factors. It could also explain observed immunological changes such as abnormalities of lymphocytes, proteins, autoantibodies, and cytokines. However, direct studies of viral infections in individuals with these psychiatric diseases have been predominantly negative. Most studies have examined antibodies in blood or cerebrospinal fluid, and relatively few studies have been done on viral antigens, genomes, cytopathic effect on cell culture, and animal transmission experiments. Viral research on schizophrenia and bipolar disorder is thus comparable to viral research on multiple sclerosis and Parkinson's disease: an attractive hypothesis with scattered interesting findings but no clear proof. The application of molecular biological techniques may allow the identification of novel infectious agents and the associations of these novel agents with serious mental diseases.