Current textbooks link influenza pandemics to influenza A virus subtypes H2 (1889-91), H3 (1990), H1 (1918-20), H2 (1957-58) and H3 (1968), a pattern suggesting subtype recycling in humans. Since H1 reappeared in 1977, whatever its origin, some workers feel that H2 is the next pandemic candidate. This report reviews the publications on which the concept of influenza A virus subtype recycling is based and concludes that the data are inconsistent with the purported sequence of events. The three influenza pandemics prior to 1957-58 were linked with subtypes through retrospective studies of sera from the elderly, or through seroarchaeology. The pandemic seroarchaeological model for subtype H1 has been validated by the recent recovery of swine virus RNA fragments from persons who died from influenza in 1918. Application of the model to pre-existing H3 antibody among the elderly links the H3 subtype to the pandemic of 1889-91, not that of 1900 as popularly quoted. Application of the model to pre-existing H2 antibody among the elderly fails to confirm that this subtype caused a pandemic in the late 1800's, a finding which is consistent with age-related excess mortality patterns during the pandemics of 1957 (H2) and 1968 (H3). H2 variants should be included in pandemic planning for a number of reasons, but not because of evidence of recycling. It is not known when the next pandemic will occur or which of the 15 (or more) haemagglutinin subtypes will be involved. Effective global surveillance remains the key to influenza preparedness.