BACKGROUND: We examined the potential clinical utility of using a cytomegalovirus (CMV)-specific T cell immunoassay to determine the risk of developing new-onset CMV retinitis (CMVR) in patients with acquired immunodeficiency syndrome (AIDS). METHODS: CMV-specific T cell assays were performed by multiparameter flow cytometry using stored peripheral blood mononuclear cells that had been obtained in an observational study 2-6 months before new-onset CMVR was diagnosed in case patients (at a study visit during which a dilated ophthalmologic examination revealed no evidence of CMVR) and at the same study visit in control subjects (matched by absolute CD4(+) T cell count at entry) who did not subsequently develop retinitis during 1-6 years of study follow-up. RESULTS: There were no significant differences in CMV-specific CD4(+) or CD8(+) T cell interferon-gamma or interleukin-2 expression in peripheral blood mononuclear cells from case patients and control subjects. Although there were trends toward lower percentages and absolute numbers of CMV-specific, cytokine-expressing CD8(+) T cells with a "late memory" phenotype (CD27(-)CD28(-)) as well as with an "early memory" phenotype (CD27(+)CD28(+)CD45RA(+)) in case patients than in control subjects, these differences were not statistically significant. CONCLUSIONS: Many studies have reported that CMV-specific CD4(+) and CD8(+) T cell responses distinguish patients with active CMVR (i.e., who lack CMV-protective immunity) from those with inactive CMVR after immune restoration by antiretroviral treatment (i.e., who have CMV-protective immunity). However, the multiple CMV-specific immune responses we measured do not appear to have clinical utility for predicting the risk for patients with AIDS of developing new-onset CMVR with sufficient accuracy to be used in guiding therapeutic management.