The immune system has a memory that it exhibits in the enhanced and augmented responses the second time it meets an antigen. The memory is the result of a number of changes to the system brought about during the primary response. The most important of these changes is the formation of an expanded pool of antigenspecific memory cells. One of the enduring questions in immunology is how these memory cells are maintained for such long periods. Until about 10 years ago, memory cells were thought to be very long-lived cells that required little, if any, external input to keep them alive. This view arose not out of data showing long cellular lifespans, but from the observation that memory responses could be elicited many years after immunization or infection. It has been a common failure in thinking about immunological memory to imbue the individual cells with properties of the whole system. Around 10–12 years ago, a number of studies were published that highlighted the influence of antigen on the survival of memory cells. These studies indicated that although the antigen-specific clones were long-lived, the constituent memory cells were relatively short-lived and required continued antigenic stimulation (1–3). In the intervening period, the argument about antigen dependence has continued, and recent experiments have swung opinion back towards an antigenindependent view. In this article, I wish to argue that in reality, as opposed to experimental models, the role of antigen in memory maintenance is a very important one, and that antigen persistence might be a crucial asset for any vaccine.