In a 7-year prospective study cytomegalovirus (CMV) was shown to infect approximately twice as many pregnant women as did rubella virus. Fetal loss occurred in 4/26 (15%) early CMV infections which was seven-fold higher than the rate found in controls (16/744; 2.2%). There was no evidence that fetal loss resulted from intrauterine transmission of virus. Fifty-eight women experienced primary CMV infection and congenital infection was found in nine (20%) of the 46 infants from whom clinical samples were obtained. Transmission of virus was found in 20%, 0% and 40% in the first, second and third trimesters respectively. All babies were normal at birth but two have so far developed definite intellectual impairment attributable to cytomegalovirus infection. The mothers of both of these cases were infected after the fetus had become legally viable. We conclude that the lessons learned from studying rubella infection during pregnancy cannot be applied to cytomegalovirus; in particular, we could find no evidence that termination of pregnancy should be offered to women with early CMV infections.