Influenza A viruses pose a serious threat to public health, and seasonal circulation of influenza viruses causes substantial morbidity and mortality. Influenza viruses continuously acquire substitutions in the surface glycoproteins hemagglutinin (HA) and neuraminidase (NA). These substitutions prevent the binding of pre-existing antibodies, allowing the virus to escape population immunity in a process known as antigenic drift. Due to antigenic drift, individuals can be repeatedly infected by antigenically distinct influenza strains over the course of their life. Antigenic drift undermines the effectiveness of our seasonal influenza vaccines and our vaccine strains must be updated on an annual basis due to antigenic changes. In order to understand antigenic drift it is essential to know the sites of antibody binding as well as the substitutions that facilitate viral escape from immunity. In this dissertation, we explore both the epitopes targeted in human antibody responses and how influenza viruses evade these responses. We first demonstrate that prior exposure shapes the sites targeted in human antibody responses, and show that many middle-age adults mounted an antibody response against H1N1 viruses that is focused against sites on HA conserved between contemporary strains and strains that circulated in early childhood. In addition, we demonstrate that a viral substitution in this epitope allows influenza viruses to evade neutralizing antibody responses. We next demonstrate that an H3N2 HA substitution introducing a glycosylation site prevents the binding of neutralizing antibodies present in a large number of individuals. Importantly, our egg-based vaccines lack this glycosylation due to culture-adaptive substitutions, but a vaccine containing this glycosylation motif more potently induced antibody responses against circulating strains. Finally, we identify and characterize antibodies that target conserved residues in the receptor-binding site (RBS) of HA. We demonstrate that in some individuals RBS antibodies in sera contribute to neutralization of antigenically distinct strains, even in the case of an antigenically mismatched vaccine. Overall, the work presented here helps address the complex interaction of influenza viruses and human immunity. Importantly, our work identifies shortcomings with our current process of vaccine strain selection and production and investigates epitopes of interest for universal influenza vaccine efforts.