This study examined the methylation difference in AIRE and RASSF1A between maternal and placental DNA, and the implication of this difference in the identification of free fetal DNA in maternal plasma and in prenatal diagnosis of trisomy 21. Maternal plasma samples were collected from 388 singleton pregnancies, and placental or chorionic villus tissues from 112 of them. Methylation-specific PCR (MSP) and methylation-sensitive restriction enzyme digestion followed by fluorescent quantitative PCR (MSRE + PCR) were employed to detect the maternal-fetal methylation difference in AIRE and RASSF1A. Diagnosis of trisomy 21 was established according to the ratio of fetal-specific AIRE to RASSF1A in maternal plasma. Both methods confirmed that AIRE and RASSF1A were hypomethylated in maternal blood cells but hypermethylated in placental or chorionic villus tissues. Moreover, the differential methylation for each locus could be seen during the whole pregnant period. The positive rates of fetal AIRE and RASSF1A in maternal plasma were found to be 78.1% and 82.1% by MSP and 94.8% and 96.9% by MSRE + PCR. MSRE + PCR was superior to MSP in the identification of fetal-specific hypermethylated sequences (P < 0.05). Based on the data from 266 euploidy pregnancies, the 95% reference interval of the fetal AIRE/RASSF1A ratio in maternal plasma was 0.33-1.77, which was taken as the reference value for determining the numbers of fetal chromosome 21 in 102 pregnancies. The accuracy rate in 98 euploidy pregnancies was 96.9% (95/98). Three of the four trisomy 21 pregnancies were confirmed with this method. It was concluded that hypermethylated AIRE and RASSF1A may serve as fetal-specific markers for the identification of fetal DNA in maternal plasma and may be used for non-invasive prenatal diagnosis of trisomy 21.