CD8 T cells are important antiviral effectors in the adaptive immune response to cytomegaloviruses (CMV). Naïve CD8 T cells can be primed by professional antigen-presenting cells (pAPCs) alternatively by "direct antigen presentation" or "antigen cross-presentation". In the case of direct antigen presentation, viral proteins are expressed in infected pAPCs and enter the classical MHC class-I (MHC-I) pathway of antigen processing and presentation of antigenic peptides. In the alternative pathway of antigen cross-presentation, viral antigenic material derived from infected cells of principally any cell type is taken up by uninfected pAPCs and eventually also fed into the MHC class-I pathway. A fundamental difference, which can be used to distinguish between these two mechanisms, is the fact that viral immune evasion proteins that interfere with the cell surface trafficking of peptide-loaded MHC-I (pMHC-I) complexes are absent in cross-presenting uninfected pAPCs. Murine cytomegalovirus (mCMV) models designed to disrupt either of the two presentation pathways revealed that both are possible in principle and can substitute each other. Overall, however, the majority of evidence has led to current opinion favoring cross-presentation as the canonical pathway. To study priming in the normal host genetically competent in both antigen presentation pathways, we took the novel approach of enhancing or inhibiting direct antigen presentation by using recombinant viruses lacking or overexpressing a key mCMV immune evasion protein. Against any prediction, the strongest CD8 T-cell response was elicited under the condition of intermediate direct antigen presentation, as it exists for wild-type virus, whereas the extremes of enhanced or inhibited direct antigen presentation resulted in an identical and weaker response. Our findings are explained by direct antigen presentation combined with a negative feedback regulation exerted by the newly primed antiviral effector CD8 T cells. This insight sheds a completely new light on the acquisition of viral immune evasion genes during virus-host co-evolution.