Independent studies have shown that CD27, 4-1BB, and OX40 can all promote survival of activated CD8+ T cells. We have therefore compared their impact on CD8+ memory T cell formation and responsiveness within one, physiologically relevant model system. Recombinant mice, selectively lacking input of one or two receptors, were challenged intranasally with influenza virus, and the immunodominant virus-specific CD8+ T cell response was quantified at priming and effector sites. Upon primary infection, CD27 and (to a lesser extent) 4-1BB made nonredundant contributions to accumulation of CD8+ virus-specific T cells in draining lymph nodes and lung, while OX40 had no effect. Interestingly though, in the memory response, accumulation of virus-specific CD8+ T cells in spleen and lung critically depended on all three receptor systems. This was explained by two observations: 1) CD27, 4-1BB, and OX40 were collectively responsible for generation of the same memory CD8+ T cell pool; 2) CD27, 4-1BB, and OX40 collectively determined the extent of secondary expansion, as shown by adoptive transfers with standardized numbers of memory cells. Surprisingly, wild-type CD8+ memory T cells expanded normally in primed OX40 ligand- or 4-1BB ligand-deflcient mice. However, when wild-type memory cells were generated in OX40 ligand- or 4-1BB ligand-deficient mice, their secondary expansion was impaired. This provides the novel concept that stimulation of CD8+ T cells by OX40 and 4-1BB ligand during priming imprints into them the capacity for secondary expansion. Our data argue that ligand on dendritic cells and/or B cells may be critical for this.