We performed a prospective, randomized trial in CMV seronegative marrow recipients to determine if filtered blood products were as effective as CMV-seronegative blood products for the prevention of transfusion- transmitted CMV infection after marrow transplant. Before transplant, 502 patients were randomized to receive either filtered or seronegative blood products. Patients were monitored for the development of CMV infection and tissue-documented CMV disease between days 21 and 100 after transplant. Infections occurring after day 21 from transplant were considered related to the transfusion of study blood products and, thus, were considered evaluable infections for the purpose of this trial. In the primary analysis of evaluable infections, there were no significant differences between the probability of CMV infection (1.3% v 2.4%, P = 1.00) or disease (0% v 2.4%, P = 1.00) between the seronegative and filtered arms, respectively, or probability of survival (P = .6). In a secondary analysis of all infections occurring from day 0 to 100 post-transplant, although the infection rates were similar, the probability of CMV disease in the filtered arm was greater (2.4% v 0% in the seronegative arm, P = .03). However, the disease rate was still within the prestudy clinically defined acceptable rate of < or = 5%. We conclude that filtration is an effective alternative to the use of seronegative blood products for prevention of transfusion- associated CMV infection in marrow transplant patients.