We propose here a stochastic model for the CD 8(+)T lymphocyte dynamics on the long time-scale of the human lifespan. Our purpose has been to test the hypothesis, recently proposed on the basis of our experimental data (Fagnoni et al., 2000), that the depletion of virgin CD8(+)T lymphocytes can be considered a reliable biomarker related to the risk of death. This hypothesis is embedded in a more general theory of immunosenescence according to which the accumulation of antigen experienced (AE) T cells and the concomitant exhaustion of antigen non-experienced (ANE) T cells with age, mostly due to the chronic lifelong exposure to antigens, is a major characteristic of the remodeling of the human immune system with age. In our model we considered a deterministic balance of ANE and AE T cell concentrations plus a stochastic forcing, which describes the chronic antigenic stress fluctuations, assuming a mean genetically determined capability of individuals to respond to antigens. The major results of our model is the validation of the above-mentioned hypothesis, since the model is capable of fitting the experimental data concerning the changes of ANE T cell concentration over age, and at the same time to reproduce survival curves similar to the demographic ones. Furthermore, the stochastic process results in being responsible for the peculiar shape of the survival curves.