It was found by the technique of molecular hybridization that the pandemic influenza virus strains of 1957 (H2N2) and 1968 (H3N2) evolved by reassortment of RNA segments from the foregoing pandemic strains, replacing four genes including those coding for haemagglutinin (HA) and neuraminidase (NA) (1957) or only that coding for the haemagglutinin (1968), respectively. The earlier pandemic strains from 1918-9 (as deduced from the swine influenza strain (Hsw1N1), which is assumed to be a survivor of the Spanish influenza), from 1933-4 (H0N1), and from 1947 (H1N1) were derived from each other through a number of point mutations only. The Russian strain from 1977 (H1N1) is genetically almost identical with the FW strain from 1950 (H1N1). In contrast to the conserved genes coding for the internal viral proteins, the genes coding for the viral surface glycoproteins consist of a relatively small highly conserved part which presumably is responsible for the functional integrity of the gene products, and the relatively large variable region which probably determines the antigenic property of their gene products. This structural feature of the surface glycoprotein genes explains the fast antigenic drift by mutations in the variable region and selection by the immune response of the host.