This trial was designed to compare foscarnet with ganciclovir as pre-emptive therapy for CMV infection in patients undergoing allogeneic hemopoietic stem cell transplant (HSCT). Thirty-nine patients were randomized to receive foscarnet 90 mg/kg every 12 h (n = 20) or ganciclovir 5 mg/kg every 12 h (n = 19) for 15 days at the time of development of CMVAg-emia. Primary-end points of the study were (1) outcome of CMVAg-emia; (2) progression to CMV disease; and (3) side-effects of treatment. The secondary end-point was transplant-related mortality (TRM). The two groups were comparable for diagnosis, status of disease, donor type, acute graft-versus-host (aGVHD) prophylaxis, interval between HSCT and CMVAg-emia and number of CMVAg positive cells; the donor and recipient age were borderline older in the foscarnet group. Increments of serum creatinine in the foscarnet group, and cytopenia in the ganciclovir group were controlled by reducing the administered dose: in the first 15 days of therapy 9/20 foscarnet and 10/19 ganciclovir patients had a dose reduction greater than 20% (P = 0.43). Clearance of CMVAg-emia was faster in the foscarnet group although with borderline statistical significance. Failures of treatment occurred in 3/20 patients in foscarnet group vs 8/19 patients in ganciclovir group (P= 0.06): causes of failure were the need for combination therapy to control antigenemia (1/20 vs 5/19), and reactivation during treatment for 2 vs 3 patients, respectively. CMV disease was diagnosed in 1 vs 2 patients (P = 0.5) who subsequently died. The actuarial 1-year TRM was 25 vs 12%, respectively (P = 0.3). This study suggests that foscarnet and ganciclovir are both effective for pre-emptive therapy of CMVAg-emia, although the number of failures would seem to be slightly higher in the ganciclovir patients. Side-effects are seen in both groups and can be managed with appropriate dose reduction.