Eight laboratory assays, viz., the pp65 direct antigenemia test, a quantitative cytomegalovirus (CMV)-specific immunoglobulin G (IgG) assay (Biomerieux VIDAS), a CMV-specific IgM assay (Biomerieux VIDAS), the Hybrid Capture system (Murex), an in-house PCR with plasma (P-PCR) and leukocytes (L-PCR), and a commercial PCR (Roche AMPLICOR) with plasma (P-AMP) and leukocytes (L-AMP), were compared for their abilities to predict CMV disease before the onset of illness in a prospective study of 37 renal transplant recipients. By using an expanded criterion for active infection (two or more of the markers positive) and a clinical definition of disease, 22 (59%) patients were identified as having active CMV infection and 13 (35%) were identified as having CMV disease. Of the 13 CMV-seronegative recipients who received seropositive kidneys (R- group), 8 had active infection and disease. All assays were 100% specific and 100% predictive of CMV disease in the R- group. The leukocyte PCRs (L-PCR and L-AMP) were the most sensitive assays, had positive results an average of between 8 and 13 days before the onset of illness, and were the assays of choice. The performance of the assays was less satisfactory for the 24 patients who were CMV seropositive before transplantation (R+ group). A negative result was more useful for this group. Overall, P-AMP had the best results, and it could be the assay of choice for monitoring R+ patients. The non-PCR-based methods generally had high specificities but often gave late positive results and were not sensitive enough for use as prediction tools for either group of patients.