Influenza causes widespread morbidity and mortality, especially during pandemics. Current strain-matched vaccines leave public health gaps due to the long lead time for strain selection, vaccine production, and distribution. Universal influenza vaccines inducing broad (heterosubtypic) immunity could protect against diverse strains and even different subtypes of virus, and they could be available off-the-shelf early in an outbreak. Broad cross-protection to influenza has long been studied in animals (reviewed in [1]). Prior immunity to other strains does not prevent infection, but it greatly reduces morbidity and mortality. Extensive literature documents the ability of conserved antigens including nucleoprotein (NP) and matrix 2 (M2) to induce cross-protection [2–7], and there is evidence that cross-protective vaccination reduces transmission [8]. Nonneutralizing antibodies may also protect [9–11] perhaps by antibody‐dependent cell‐mediated cytotoxicity [9, 11]. Cross-protection in humans is more controversial. Human T-cell memory responses to conserved influenza antigens are well documented, and they cross-react with virus strains the individuals have never encountered [12–14]. However, there is skepticism about whether T cells protect against disease [15]. Linking immunity to protection requires challenge or surveillance studies, and such evidence has accumulated gradually. A classic human study found that, in donors lacking neutralizing antibodies, pre-existing cytotoxic T-cell levels against influenza correlated with reduced virus shedding but not reduced symptoms after influenza virus challenge [16]. Epidemiological studies of the 1957 pandemic, an H1N1 to H2N2 shift, suggested