The mechanistic basis of memory T-cell development is poorly defined. Phenotypic markers that define precursors at effector stages have been characterized for acute systemic infections with high antigen load. We asked whether such markers can identify memory precursors from early effectors (d6) to late memory (>d500) for two immunodominant CD8(+) responses during the course of a localized low-load influenza infection in mice. CD8(+) T cells stained with the D(b) NP(366) and D(b) PA(224) tetramers were characterized as IL-7Rα(hi) , IL-7Rα(hi) CD62L(hi) or IL-7Rα(hi) KLRG1(lo) . While the D(b) NP(366) - and D(b) PA(224) -specific responses were comparable in size, decay kinetics and memory precursor frequency, their expansion characteristics differed. This correlated with a divergence in the IL-7Rα(hi) , IL-7Rα(hi) CD62L(hi) and IL-7Rα(hi) KLRG1(lo) phenotypes on effector, but not naïve, CD8(+) populations. That effect was abrogated by priming with viruses engineered to present equivalent levels of NP(366) and PA(224) peptides, indicating that memory phenotypes reflect early antigenic experience rather than memory potential. Thus, the IL-7Rα(hi) KLRG1(lo) phenotype had a poor predictive value in identifying memory precursors in the spleen and at the site of infection. Greater consistency in influenza-specific IL-7Rα(hi) KLRG1(lo) CD8(+) T-cell numbers was found in draining lymph nodes, suggesting that this may be the preferential site for memory establishment and maintenance following localized virus infections.