Objectives: To derive models for estimating risk of miscarriage and stillbirth from maternal characteristics and findings of first-trimester screening for aneuploidies and to define the procedure-related risk of chorionic villus sampling (CVS) after adjusting for these factors. Method: We examined 33 856 singleton pregnancies at 11+0 to 13+6 weeks, and in 2396 CVS was carried out. Logistic regression analysis was used to examine the factors contributing to miscarriage and stillbirth. Results: There were 33 310 (98.4%) livebirths, 404 (1.2%) miscarriages and 142 (0.4%) stillbirths. Models combining maternal characteristics, nuchal translucency, pregnancy-associated plasma protein-A (PAPP-A) and flow in the ductus venosus detected 36.9% of miscarriages and 35.2% of stillbirths, at a 10% false-positive rate. The risk of miscarriage and stillbirth increased with maternal age and weight, in women of African racial origin, in those with previous miscarriages or stillbirths and in those with low serum PAPP-A and reversed A-wave in the ductus venosus. The risk of miscarriage increased in women with pre-existing diabetes mellitus, in those conceiving on ovulation-induction drugs and in those with high fetal nuchal translucency, and the risk of stillbirth increased in women with chronic hypertension and in cigarette smokers. The expected number of miscarriages and stillbirths in the CVS group and the models derived from the non-CVS group were 45 (95% prediction intervals 32-58) and 18 (95% prediction intervals 9-26), respectively. These expected numbers were not significantly different from the observed 44 and 15 cases (p = 0.881 and 0.480), respectively. Conclusion: A high proportion of fetal losses can be predicted at 11 to 13 weeks. A model for such predictions can be used to assess the procedure-related risks from CVS. Copyright © 2011 John Wiley & Sons, Ltd.