In contrast to most organs, the anatomy of the liver may allow naive CD8⁺ T cells to make direct contact with liver parenchymal cells. We have previously shown, using a combination of TCR transgenic T cells specific for H-2 K^b and hepatocytes expressing a transgenic H-2 K^b molecule, that hepatocytes can induce antigen-specific activation and proliferation of naive CD8⁺ T cells independently of CD28 co-stimulation. However, T cell activation by hepatocytes leads to premature T cell death and tolerance, bothin vivo andin vitro. In this study, we investigated the mechanisms of T cell death induced by hepatocytesin vitro using primary hepatocytes to activate purified CD8⁺ T cells. Neither Fas nor tumor necrosis factor receptor were involved, indicating that hepatocyte- induced death was distinct from activation-induced cell death. Before they started to divide, T cells activated by hepatocytes expressed lower levels of thebcl-x_L survival gene and 30 times less IL-2 mRNA than CD8⁺ cells activated by splenic antigen-presenting cells. Since CD28 co-stimulation increases both IL-2 andbcl-x_L expression, this suggests that hepatocyte-activated T cells die by neglect because they fail to receive effective co-stimulatory signals. In agreement with this model, premature death promoted by hepatocytes could be prevented by cross-linking CD28. Survival after CD28 cross-linking correlated with increased IL-2 andbcl-x_L expression, and sustained T cell proliferation, while cytotoxic T lymphocyte activity was prolonged as compared with cells stimulated without CD28 co-stimulation. This study confirms that high- affinity TCR transgenic antigen-specific CD8⁺ T cells can be activated to proliferate and differentiate into cytotoxic effector cells. However, prolonged T cell survival and cytotoxicity required CD28 co-stimulation as well. To our knowledge, this is the first report suggesting that tolerance in the context of lack of CD28 co-stimulation can result from Fas-independent peripheral deletion rather than from anergy.