There is still uncertainty on the correlates of protection by influenza vaccine. To determine the relationship between hemagglutination-inhibition (HI) titer and the specificity and avidity of serum antibodies, we analyzed serum from a longitudinal trial (1983-1987) of influenza vaccine efficacy [Keitel WA, Cate TR, Couch RB, Huggins LL, Hess KR. Efficacy of repeated annual immunization with inactivated influenza virus vaccines over a five year period. Vaccine 1997;15(10):1114-22 ]. We captured native virus particles with fetuin and separately measured relative antibody levels and avidities of antibodies against native glycoproteins and antibodies against denatured viral proteins. Most subjects had pre-existing antibodies against A/Victoria/75 and, although 70% had >two-fold increased antibodies against A/Philippines/82 after vaccination, only 30% showed increased antibodies to A/Victoria/75 indicating no dominance of original antigenic sin. There was variation in the levels of antibodies to unfolded antigens compared to native, but antibodies against denatured proteins never exceeded those against native virus. In some cases, the avidity increased without a significant increase in antibody concentration, which might explain why some vaccinees with low HI titer demonstrate adequate protection. We found that the negative correlation between pre-vaccination HI titer and the increase after vaccination is also seen when antibodies are measured directly, but that there is little relationship between HI titer and antibodies against native glycoproteins, either in amount or avidity. Our assay, which has also been adapted for recent influenza viruses that do not bind to fetuin, may be useful for vaccine evaluation.