The current vaccination strategy against influenza A and B viruses is vulnerable to the unanticipated emergence of epidemic strains that are poorly matched by the vaccine. A vaccine that is less sensitive to the antigenic evolution of the virus would be a major improvement. The general feasibility of this goal is supported by studies in animal models that show that immunologic activities directed against relatively invariant viral determinants can reduce illness and death. The most promising approaches are based on antibodies specific for the relatively conserved ectodomain of matrix protein 2 and the intersubunit region of hemagglutinin. However, additional conserved determinants for protective antibodies are likely to exist, and their identification should be encouraged. Most importantly, infection and current vaccines do not appear to effectively induce these antibodies in humans. This finding provides a powerful rationale for testing the protective activity of these relatively conserved viral components in humans.