Objective To investigate the tendency and safety of percutaneous umbilical cord blood sampling (PUBS) in prenatal screening and diagnosis, and the possibilities of avoiding unnecessary PUBS. Methods This was a retrospective study of pregnant women who underwent PUBS for prenatal diagnosis in Peking University First Hospital from January 2015 to December 2017. Clinical indications, timing of PUBS, further investigations (chromosome karyotype, molecular genetics and pathogen testing), results, and pregnancy outcomes were collected and analyzed. One-way analysis of variance (ANOVA), Chi-square test for linear trend, Fisher's exact probability test/Cochran-Armitage analysis and Cruskal-Wallis rank-sum test were used for statistical analysis. Results (1) A total of 412 singleton pregnancies underwent PUBS at 20-38 gestational weeks during the study period, and 379 (92.2%) of them received PUBS before 34 gestational weeks. The positive test results accounted for 10.4% (43/412). There were six (1.5%) miscarriages after PUBS. In vitro cell culture failure occurred in two cases, one in 2015 and the other in 2016. (2) Among the 412 cases, 304 (73.8%) had only one indication. Fourteen cases could be identified as high risk in the first trimester, such as advanced maternal age (AMA, >35 years), pregnant history with chromosomal abnormal fetus and one of the couples carrying abnormal genes. There were four, zero and one case receiving PUBS only for AMA in 2015, 2016 and 2017, respectively. Indications, including high risk suggested by serum screening and fetal abnormality found by ultrasound were identified in 290 cases (70.4%) in the second or third trimester. Other than AMA, there were no statistically significant differences in single indicators. The proportion PUBS with double indicators increased from 2015 to 2017 but without significant difference. AMA and positive serum screening as indicators of aneuploidy screening accounted for 7.6% (8/105) in double-indicator group and 1.9% (8/412) in all. (3) There were 363 PUBS (88.1%) performed for ultrasound abnormalities. Among them, 76.9% (280/363) only had abnormal ultrasound findings, and the percentage was decreased year by year. The other 83 cases (80 with double indicators and three with triple indicators) also presented with other indicators, including AMA, adverse pregnancy history and positive serum screening. The proportion of PUBS performed with the presence of multiple indicators tended to increase recently, but no statistically significant difference was found. All the 18 cases with abnormality diagnosed by molecular genetic testing had abnormal ultrasound findings. Conclusions Although PUBS's complications are rare, it carries some risks. The constitution of single indication has been declined every year. With the improvement of prenatal screening system and application of molecular karyotyping, the necessity of invasive prenatal diagnosis with PUBS is greatly reduced. An improvement in reasonable and standardized application of PUBS needs to be achieved. Key words: Prenatal diagnosis; Phlebotomy; Blood specimen collection; Umbilical veins