Objective: To compare patient survival, graft survival and financial costs at 3 years after renal transplant in graft recipients who received pre-emptive or deferred treatment of cytomegalovirus (CMV) disease. Clinical outcomes and CMV-related charges at an average of 16 months' follow-up have been reported previously. Design and Setting: This was a randomised, nonblind trial in patients enrolled between June and December 1994 at a single US centre. Patients and Participants: 36 adult renal transplant recipients who themselves, or their donors, were CMV-seropositive were randomised into 15 patients in the pre-emptive treatment group and 21 patients in the deferred treatment group. Methods: Patients in the pre-emptive treatment group received intravenous ganciclovir 5 mg/kg every 12 hours once CMV viraemia was detected. Patients in the deferred treatment group received the same intravenous ganciclovir regimen after developing symptomatic CMV disease. Cost data were collected for inpatient hospitalisations, outpatient intravenous medications and dialysis treatments. Results: A comparison of the descriptive characteristics of the two groups revealed no significant differences. Three of 15 patients in the pre-emptive group died with graft function during the third year post-transplant, whereas none of 21 patients in the deferred group died during the study. However, none of the 3 deaths appear to be causally related to the study (myocardial infarction, lung cancer and homicide). Estimates of graft failure, censored for death with function, show a trend towards more graft loss in the deferred treatment group (p = 0.089). Average treatment costs were not significantly different during the first year post-transplant. However, by the third year post-transplant, costs were significantly higher in the deferred group (an additional $US15,277/patient, p < 0.001). Conclusions: With an average of 16 months' follow-up and an assessment of CMV-related charges, it was concluded that the deferred strategy was preferred because of equivalent clinical outcomes at lower CMV treatment charges. The results of the current study with 3-year follow-up and an analysis of total post-transplant treatment costs reverses the previous conclusion. We now conclude that a pre-emptive strategy for treating CMV disease is preferred because of similar clinical outcomes with lower total medical costs.