Medical interest in cytomegalovirus (CMV) is based on lifelong neurological sequelae, such as sensorineural hearing loss and mental retardation, resulting from congenital infection of the fetus in utero, as well as on CMV disease with multiple organ manifestations and graft loss in recipients of hematopoietic cell transplantation or solid organ transplantation. CMV infection of transplantation recipients occurs consequent to reactivation of virus harbored in a latent state in the transplanted donor cells and tissues, or in the tissues of the transplantation recipient herself or himself. Hence, CMV infection is a paradigm for a viral infection that causes disease primarily in the immunocompromised host, while infection of the immunocompetent host is associated with only mild and nonspecific symptoms so that it usually goes unnoticed. Thus, CMV is kept under strict immune surveillance. These medical facts are in apparent conflict with the notion that CMVs in general, human CMV as well as animal CMVs, are masters of 'immune evasion', which during virus-host co-speciation have convergently evolved sophisticated mechanisms to avoid their recognition by innate and adaptive immunity of their respective host species, with viral genes apparently dedicated to serve just this purpose (Reddehase in Nat Rev Immunol 2:831-844, 2002). With focus on viral interference with antigen presentation to CD8 T cells in the preclinical model of murine CMV infection, we try here to shed some more light on the in vivo balance between host immune surveillance of CMV infection and viral 'immune evasion' strategies.