OBJECTIVE: Over the past 3 years, reports of DNA alteration in myotonic dystrophy, fragile X syndrome (types A and E), Kennedy's disease, Huntington's disease, spinocerebellar ataxia type 1, and dentatorubral- pallidoluysian atrophy have identified a new class of human mutation, referred to as trinucleotide repeat amplification. All available evidence suggests that this unstable trinucleotide repeat DNA is the biological basis of the clinical phenomenon of genetic anticipation. Two components of anticipation, greater severity and earlier age at onset in subsequent generations, have been widely observed in schizophrenia and bipolar affective disorder. Thus, a reanalysis of the genetics of major psychosis from the perspective of unstable DNA is of significant interest. METHOD: The authors reviewed the available literature on anticipation and related phenomena in major psychosis and reevaluated the family, twin, and adoption study data. RESULTS: The unstable DNA concept competes well with the traditional multifactorial polygenic theory; many deviations from a single gene mode of inheritance in psychiatric twin and family studies, which previously served as strong proof for more than one etiologic gene, can be easily explained by the non-Mendelian behavior of unstable DNA. In addition, this new paradigm provides a simple explanation for unclear issues in the genetics of major psychosis, such as the identical rate of psychosis in the offspring of discordant monozygotic twins. CONCLUSIONS: The major advantage of the unstable DNA hypothesis over the multifactorial polygenic theory lies in the possibility of falsifying the unstable DNA hypothesis by two independent laboratory strategies: a classical linkage analysis and a set of novel methods for the direct detection of unstable DNA sites.Abstract Teaser