Résumé : It has been difficult to measure efficacy of T cell-based vaccines and to correlate efficacy of CD8$^+$T cell responses with protection against viral infections. In part, this difficulty is due to our poor understanding of the in vivo efficacy of CD8$^+$T cells. Using a recently developed experimental method of in vivo cytotoxicity we investigated quantitative aspects of killing of peptide-pulsed targets by effector and memory CD8$^+$T cells, specific to three epitopes of lymphocytic choriomeningitis virus (LCMV), in the mouse spleen. By analyzing data on killing of targets with varying number of epitope-specific effector and memory CD8$^+$T cells, we find that killing of targets by effectors follows the law of mass-action, that is the death rate of peptide-pulsed targets is proportional to the frequency of CTLs in the spleen. In contrast, killing of targets by memory CD8$^+$T cells does not follow the mass action law because the death rate of targets saturates at high frequencies of memory CD8$^+$T cells. For both effector and memory cells, we also find no support for a killing term that includes the decrease of the death rate of targets with increasing target cell density. Importantly, we find that at low CD8$^+$T cell frequencies, effector and memory CD8$^+$T cells on the per capita basis are equally efficient at killing peptide-pulsed targets. Our framework provides the guideline for the calculation of the level of memory CD8$^+$T cells required to provide sterilizing protection against viral infection. Our results thus form a basis for quantitative understanding of the process of killing of virus-infected cells by T cell responses in tissues and can be used to correlate the phenotype of vaccine-induced memory CD8 T cells with their killing efficacy in vivo.