ALTHOUGH some viruses, particularly the herpes viruses, may never be eliminated from the body¹, others like influenza A, regularly reinfect humans and boost waning crossreactive CD8⁺ T-cell immunity². Prolonged T-cell memory is found for viruses that are unlikely to be re-encountered and which do not persist in the host genome^3–5, indicating that CD8⁺ T-cell memory might be independent of continued (or sporadic) antigenic exposure. A feature of virus-specific CD8⁺ T-cell memory is that antigen-specific cytotoxic T-lymphocyte precursors (CTLp) are greatly increased and remain high throughout life. The idea that persistence of the inducing antigen is essential is based on experiments in which adoptively transferred CD8⁺ memory T cells could not be detected for more than a few weeks in naive recipient mice without secondary challenge^{6, 7}. Here we show that restimulation of such chimaeric mice with an inducing Sendai virus antigen increases the clonal burst size more than 7-fold within 8 days, making memory CTLp easier to detect in the longer term. We find that Sendai-virus-specific CTLp are maintained for >250 days in irradiated unin-fected recipients, including reconstituted β₂-microglobulin^-/-mice. To determine whether a source of viral peptide can persist after primary infection, we gave Sendai-virus-specific Thyl.l⁺memory spleen cells to naive mice that had been minimally depleted of Thy 1.2⁺ T cells, or to comparable recipients that had recovered from infection with Sendai virus or influenza virus. Although antibody against Sendai virus was never found in the naive recipients, Sendai-virus-specific CD8⁺ memory T cells were maintained equally well in each case for >100 days after cell transfer. We find no evidence for persisting depots of viral protein that might feed into the endogenous processing pathway and maintain viruspecific CD8⁺ T-cell memory.