It is not fully understood how antigenic drift of the haemagglutinin of type A influenza virus in man occurs in the presence of the expected polyclonal antibody response to the five antigenic sites, A to E. Here we show that 12 % (11/92) of sera from mice which had mounted a secondary immune response to inactivated influenza virus were able to select escape mutants. No escape mutant was selected with serum from nonimmunized mice (0/65). Selection required only a single passage, and escape mutants were identified by their reaction with monoclonal antibodies (MAbs); all but one had altered reactivity at site A. Most of the site A escape mutants (7/10) were conventional in character and did not react in haemagglutination-inhibition (HI) or neutralization assays with the identifying MAb. The HA genes of three of these were part sequenced and had a predicted single amino acid substitution (Gly-144 → Glu) in site A. The other escape mutants (3/10) had a small (2-fold) reduction in HI and neutralization to the site A MAb, but no amino acid substitution in site A. The final mutant was a conventional site B escape mutant. To model antisera which selected escape mutants, we constructed ‘pseudo-immune sera’ using mixtures of two neutralizing MAbs in which the first MAb was held at a constant high concentration (1000 HIU/ml). Escape mutants could be selected to the first MAb when the titre of the second MAb was reduced to a low but still inhibiting concentration (1 to 3 HIU/ml). Mixtures of three MAbs also selected escape mutants with similar facility provided that the second and third MAbs were reduced to a similar low concentration. Thus it is possible that the ability of an antiserum to select escape mutants is due to the neutralizing antibody response being biased to an epitope/cross-reacting epitopes within a single antigenic site. However, when escape mutants were reacted in HI assay with their selecting antiserum, the maximum difference from the titre with wt virus was 75 %. The findings of this study may be relevant to the understanding of antigenic drift in type A human influenza virus, and to immune-driven antigenic variation in other virus infections.