Serological cross-reactions between influenza viruses of the two major types A and B and between the viruses and extracts of host tissue were investigated to elucidate the nature and source of the antigens concerned. The results render untenable the generally accepted view that virus types A and B are totally unrelated antigenically and that any cross-reactivity which occurs must be due to impurities in the virus preparations used. Strong cross-complement fixation reactions were consistently demonstrable with highly purified viruses provided that the antisera were prepared by immunizing rabbits with heat-degraded viruses in which the specific antigens had been destroyed. Cross-reactions between the viruses and chorioallantoic membrane extracts showed that the virus-group antigens differed from the dominant membrane antigen which is of the Forssman type. The cross-reacting antigens of the two viruses were also found to be distinguishable from each other by their different responses to heat degradation. With the classical strains PR8 and Lee propagated in embryonated eggs, the virus-group antigen could not be removed by successive cycles of virus purification. The method of purification was adsorption of virus on human group O red cells followed by spontaneous elution effected by the receptor-destroying enzyme of the virus. The ratio of specific antigen to group antigen remained constant through six successive purification cycles. It is suggested that the host-tissue component of the viruses is an integral part of their structure which may be essential for virus synthesis and may have some determinant function in respect of virus behaviour. The possibility of its fortuitous inclusion within the virus matrix without serving any biological purpose has not been excluded, but in such case its presence would point to a replication mechanism different from simple binary fission.