The leukocyte common antigen isoforms CD45RA and CD45RO have long been used to discriminate human naive and memory T cells respectively. This model was largely based on the observation that CD45RO+ T cells respond preferentially to and show a higher frequency of precursors specific for recall antigens. However, CD45RA+ T cells have more stringent requirements for stimulation and standard in vitro assays may favour CD45RO+ cells in this respect. We tested the hypothesis that CD45RA+ T cells respond poorly to in vitro stimulation with recall antigens because of inadequate stimulation rather than a lack of precursors. Limiting dilution analyses (LDA) for tetanus toxoid (TT)-specific T cells were performed in the presence or absence of exogenous anti-CD28 antibody. Addition of anti-CD28 yielded no proliferation in the absence of specific antigen. The precursor frequency for TT in the CD4+ CD45RO+ population was approximately 1:4000, while the frequency of CD4+ CD45RA+ T cells specific for TT was 4- to > 20-fold lower. Addition of anti-CD28 antibody did not significantly alter the apparent precursor frequency for CD45RO+ cells but yielded an enhancement of the value for CD45RA+ cells by 3- to > 5-fold. No enhancement of antigen-specific proliferation by anti-CD28 was observed with CD45RA+ T cells derived from cord blood, although phytohemagglutinin responses of these cells were amplified by CD28 antibody. These results indicate that conventional LDA underestimate the true precursor frequency of antigen-specific cells within the adult CD45RA+ population and support the possibility that a small number of cells revert from a primed (CD45RO+) to an unprimed (CD45RA+) state. The majority of memory T cells, however, appear to reside in the CD45RO+ population.