Haplotype-based assay has been proved efficient in noninvasive prenatal testing for various monogenic disorders in singleton pregnancies. However, its application in twin pregnancies is still blank. Here we provide a novel algorithmic approach to noninvasively assess fetal genotypes in a dizygotic twin pregnancy at risk for Duchenne muscular dystrophy (DMD). One pregnant woman carrying a dizygotic twin gestation was recruited as she was a heterozygote of DMD gene duplication and has delivered an affected son. Construction of parental haplotypes was achieved by target sequencing of DNA samples from the parent and the proband. Single nucleotide polymorphisms within target regions were classified into six categories according to parental haplotypes. Individual fetal fractions were calculated using paternal heterozygous. Maternal-inherited haplotype was deduced using relative haplotype dosage through a two-step Bayesian model. One male fetus with a lower fetal fraction of 4.6% and one female fetus with a higher fetal fraction of 10.1% were observed. The male fetus was predicted to be a DMD pathogenic variant carrier, while the female fetus was predicted to be homozygous normal. Noninvasive prenatal testing (NIPT) results were concordant with the findings of invasive prenatal diagnosis. This study is the first report of the use of NIPT for the assessment of DMD in a twin pregnancy. The algorithm provided could expand the use of NIPT to monogenic disorders in dizygotic twin pregnancies.