Abstract Background Noninvasive prenatal screening (NIPS) based on cell-free fetal DNA (cffDNA) has rapidly been applied into clinic. However, the reliability of this method largely depends on the concentration of cffDNA in the maternal plasma. The chance of test failure results or false negative results would increase when cffDNA fraction is low. In this study, we set out to develop a method to enrich the cffDNA for NIPS based on the size difference between cell-free DNA (cfDNA) of fetal origin and maternal origin, and to evaluate whether the new NIPS method can improve the test quality. Methods We utilized 10,000 previous NIPS data to optimize a size-selection strategy for enrichment. Then, we retrospectively performed our new NIPS method with cffDNA enrichment on the 1415 NIPS samples, including 1404 routine cases and 11 false negative cases, and compared the results to the original NIPS results. Results The 10,000 NIPS data revealed the fetal fraction in short cfDNA fragments (< 160 bp) is significantly higher. By using our new NIPS strategy on the 1404 routine cases, the fetal fraction increased from 11.3 ± 4.2 to 22.6 ± 6.6%, and the new method performed a significant decrease of test-failure rate (0.1% vs 0.7%, P < 0.01). Moreover, in 45.5% (5/11) of the false negative cases, fetal trisomies were successfully detected by our new NIPS method. Conclusions We developed an effective method to enrich cffDNA for NIPS, which shows an increased success rate and a reduced chance of false negative comparing to the ordinary NIPS method.