This study aims to determine if second trimester amniocentesis in twin pregnancies provides a significant independent contribution in the prediction of miscarriage or fetal loss at any stage of pregnancy. This was a retrospective cohort study of women with twin gestations booked for routine prenatal care in four fetal medicine units in Poland in the years 2010-2020. The study population included: (1) twin pregnancies that underwent amniocentesis at 16-20 weeks' gestation; (2) twin pregnancies that did not require any further testing and were followed-up routinely. Univariable and multivariable regression analysis was used to define which maternal and pregnancy characteristics provided a significant independent contribution in the prediction of miscarriage and fetal loss at any stage of pregnancy. In the study period, 2645 twin pregnancies were eligible for analysis. There were 144 cases of miscarriage defined as fetal loss of one or both twins before 24 weeks and 40 cases of intrauterine death of one or both twins after 24 weeks. A total number of 162 twin pregnancies underwent amniocentesis at 16-20 weeks' gestation. The rate of miscarriage before 24 weeks and the rate of fetal loss at any stage of pregnancy in the group that underwent amniocentesis was 10.49% and 13.58%, respectively, compared to 5.11% and 6.52% that did not undergo amniocentesis. Multivariable regression analysis showed that factors providing a significant independent contribution in the prediction of miscarriage and fetal loss at any stage of pregnancy were monochorionicity (MC), large intertwin discordance in crown-rump length (CRL), low Pregnancy Related Plasma Protein (PAPP-A) MoM and nuchal translucency (NT) above 95th centile. Amniocentesis in twin pregnancies does not provide a significant contribution in the prediction of miscarriage or fetal loss at any stage of pregnancy.