Evidence is accumulating that old individuals are more susceptible to infection with organisms to which they were previously immune, indicating that there might be a limit to the persistence of immune memory. The prevailing concept is that defects in memory T-cell populations result from inexorable end-stage differentiation as a result of repeated lifelong antigenic challenge. We discuss here mechanisms that might constrain the persistence of memory T cells and consider whether humans will suffer from memory T-cell exhaustion as life expectancy increases.