Prenatal diagnosis and carrier detection in the hemophilias have received much attention in recent years. The error rate in prenatal diagnosis by fetoscopy is less than 1%; fetoscopy is not possible, however, until the second trimester of pregnancy. Carrier detection based on bioassays of plasma has an irreducible error rate (approximately 5%?), because of the “lyonization” phenomenon in heterozygous women, and the final results are always probabilistic. New DNA methods promise to alleviate these difficulties. Prenatal diagnosis can be accomplished in the first trimester. “Lyonization” is bypassed in carrier detection, and the results may sometimes be essentially nonprobabilistic. But the DNA methods have certain limitations of their own which are not widely appreciated. Aside from cost and the necessity to adopt a new technology, there are inherent genetic problems: mothers must be heterozygous for both a disease gene and a marker gene, final results are probabilistic if the marker gene lies outside the disease gene, and multiple marker genes are often in linkage disequilibrium. We have concluded that a clinical unit planning to use the DNA methods must also maintain the conventional methods at a high level of performance.