We performed genetic analyses for the prenatal diagnosis of Duchenne's muscular dystrophy and detection of the carrier state in five families with seven pregnancies at risk for the disease. As genetic markers for the disorder, we used DNA-sequence polymorphisms detected with 12 different DNA probes derived from the vicinity of the Duchenne's muscular dystrophy locus or from within the gene, on the X chromosome. One male fetus of a proved carrier mother was predicted to be unaffected, and this was confirmed after birth. Another male fetus was predicted to be unaffected (probability, 95 percent or greater), although a crossover event had been identified in a region of the X chromosome thought to be distal to the Duchenne gene. Unfortunately, an elevated serum creatine kinase level after birth indicated that the infant had inherited the Duchenne mutation. Three male fetuses predicted to be affected with 66 percent or 95 percent probabilities were aborted, and the presence of the DNA-marker alleles was confirmed in fetal tissues. In one family, in which the maternal grandparents were unavailable, the initial genetic interpretation had to be revised after a second male fetus was analyzed with intragenic probes. Our experience suggests that despite the large number of intragenic and flanking DNA polymorphisms available, uncertainties often remain in the prenatal diagnosis of Duchenne's muscular dystrophy. Pitfalls are presented by the large size of the region in which Duchenne's mutations can occur. Crossover events in this region, which result in an exchange of DNA between two X chromosomes, can render DNA-marker studies inaccurate. Also, an autosomal recessive mutation can produce the same clinical picture.