Naive T cells are recruited into any given host response by recognizing a spectrum of possible antigens with 'sufficient' avidity. Does selecting a more functionally diverse array give better immune control? Perhaps low avidity 'killers' that 'kiss and run' operate optimally to eliminate virus-infected targets, while high avidity 'helpers' that stay faithfully in place produce more cytokine. Recent findings indeed suggest that the selection of a broad T cell receptor repertoire is characteristic of the initial phase following antigen contact, while continued exposure leads to further cycles of division and the progressive numerical dominance of 'best-fit' clonotypes. Here, we review recent advances demonstrating a link between T cell repertoire diversity and immunity to infection, and consider the potential mechanisms at play.