Objective: To compare the outcomes of the first 50 cordocenteses at mid-gestation performed by operators who either or not received training on an in vitro model earlier. Materials and Methods: Our unit keeps a prospective database on procedure details and outcomes of cordocenteses. We compared 50 consecutive procedures, from the first one onwards, performed by 5 specialists early in their maternal fetal medicine (MFM) practice, either systematically trained (study group) or not (control group) on an in vitro cordocentesis training model (n = 500 procedures). This training was introduced at the time point MFM became a formal subspeciality in Thailand. Multiple pregnancies and fetuses with chromosomal or structural abnormalities were excluded. The main outcome measures included mean duration of the procedure, the success and fetal loss rate. Results: The average duration in the control group was significantly longer than that in the study group (13.2 vs 6.4 min, p < 0.001). Conversely, the success rate was significantly lower (94.8 vs 98.8%, p = 0.011). There were no differences in fetal loss rate. Conclusion: Systematic training on a cordocentesis model reduces the time required to successfully obtain a fetal cord blood sample. Copyright © 2010 John Wiley & Sons, Ltd.