Objective-To evaluate universal antenatal screening for haemoglobinopathies. Setting-District general hospital serving a London borough with 45% ethnic minorities. Methods-Retrospective cohort study of 1444 women referred in 1688 pregnancies and 95 tertiary referrals during 101 pregnancies. Results-Unselected women at risk for sickle cell disease booked 2.7 weeks (95% confidence interval (CI) 0.14 to 5.1) later in gestation than those at risk for thalassaemia were less likely to attend counselling (83% v 93%, relative risk (RR) 0.89; 95% CI 0.85 to 0.94), their partners were less likely to be tested (77% v 95%, RR 0.81; 0.77 to 0.83), and they were less likely to accept prenatal diagnosis (22% v 90%, RR 0.37; 0.24 to 0.57).Over 99% of tertiary referrals attended counselling and had their partners tested. There were no significant diVerences in acceptance of prenatal diagnosis between those at risk of sickle cell disease and thalassaemia (55% v 67%).Unselected women at risk of sickle cell disease were significantly less likely to have their partner tested or to accept prenatal diagnosis than tertiary referrals, but not those at risk of thalassaemia. 80% of thalassaemia and 16% of SS births were prevented. Conclusions-Uptake of prenatal diagnosis among unselected women at risk of thalassaemia is similar to that reported by tertiary centres. It is considerably lower for sickle cell disease but could increase considerably if screening occurred earlier in gestation.Acceptance of counselling is universally high, suggesting that informed choices are made, and indicating a need to measure these outcomes for cost eVectiveness studies. (J Med Screen 1999;6:3-10)