During the period from 1978 to 1983, 92 pregnancies have been evaluated by fetoscopy for the prenatal diagnosis of hemophilia A. Satisfactory fetal plasma samples were obtained in 80 instances and the diagnosis-- or exclusion--of hemophilia was made by immunoradiometric assay of the factor VIII coagulant protein (VIII:CAg). The accuracy of the diagnosis established by fetoscopy has been verified after delivery or termination, and there have been no misdiagnoses resulting from laboratory error. Additional evidence for the accuracy of the analysis was the observation that the frequency of hemophilia in pregnancies of obligate carriers of the hemophilia gene, and of women whose plasma assays were indicative of the carrier state, was 29 of 59 fetuses at risk. In one case of cross-reacting material-positive hemophilia, samples obtained at fetoscopy and from the newborn infant had normal VIII:CAg levels but the infant had decreased factor VIII procoagulant activity. There were five fetal deaths resulting from fetoscopy in 55 pregnancies not intentionally terminated. Although only a small percentage of pregnant hemophilia carriers in the United States have elected to undergo fetoscopy for prenatal diagnosis, this procedure has allowed a number of pregnancies to go to term with delivery of normal males in families that were not willing to accept the risk of a hemophilic child. In eight instances, fetoscopic evaluation was sought for two successive pregnancies.