The accuracy of DNA-based prenatal diagnosis of Duchenne muscular dystrophy (DMD) was determined by study of 174 families. Only 60% of families had a living affected male, and 63% had history of a single affected male. Prenatal diagnosis was declined by 47% of mothers whose DNA studies predicted a carrier risk below 2%, and none have had affected sons. Fetal risk was estimated prospectively by linkage analysis using intragenic and flanking RFLPs and retrospectively using dystrophin cDNA analysis for families whose linkage estimates lacked precision. Diagnostic accuracy was determined by comparing predictions with 40 male pregnancy outcomes. On the basis of linkage analysis, we anticipated 3.2 DMD males and observed 3.0. Retrospective cDNA analysis identified deletions in 2 of these 3 males. The combined use of linkage and cDNA deletion analysis provided a highly accurate method for prenatal diagnosis of DMD.