About 0.5% of infants are infected in utero with cytomegalovirus (CMV). As many as 20% become mentally retarded and a further substantial proportion suffer lesser degrees of brain damage. The need for a vaccine is greater than in the case of rubella. A live, tissue culture-adapted strain of CMV has been shown to produce neutralizing antibody in volunteers without significant side effects or detectable virus excretion. The problems of developing such a vaccine for use in man, namely, attenuation, antigenic differences among virus strains, reactivation of latent infection and oncogenicity, are briefly discussed.