The aims of this study were to determine whether assumptions used in prenatal screening for Down syndrome in twin pregnancies are valid and derive estimates of risk and screening performance in twin pregnancies using observed data. Data were collected on nuchal translucency, chorionicity, pregnancy associated plasma protein‐A (PAPP‐A), and free ß human chorionic gonadotrophin (free ß‐hCG) from 61 twin pregnancies with Down syndrome and 7302 unaffected twin pregnancies. Distribution parameters were determined and used to estimate screening performance. The assumption that proportional differences in serum marker levels in affected and unaffected singleton pregnancies apply to twin pregnancies was not confirmed. Median free β‐hCG value in monochorionic affected twin pregnancies (2.63 multiples of the median [MoM]; 95% CI, 1.79‐3.22 MoM) was lower than that assuming proportionality (3.76 MoM), and the median PAPP‐A value in dichorionic affected twin pregnancies (1.88 MoM; 95% CI, 1.60‐2.17 MoM) was higher than that based on proportionality (1.33 MoM). The detection rate was 87% for a 3% false‐positive rate in monochorionic twin pregnancies and 74% in dichorionic twin pregnancies compared with 86% in singleton pregnancies. Estimates of screening performance in Down syndrome twin pregnancies do not need to rely on assumptions and can take account of chorionicity and gestational age.