Over a 2-year period from January 1991 to December 1992, second-trimester maternal serum screening for Down's syndrome using alpha-fetoprotein (alpha FP), human chorionic gonadotrophin (hCG), and unconjugated oestriol (uE3) was made available to five health districts in East Anglia, with a total population of 1.2 million. Amniocentesis was offered when the risk of Down's syndrome at term was 1:200 or greater. 25,359 singleton pregnancies were screened, representing an uptake of 77 per cent. The recall rate for the 24 per cent of women who had not had a dating scan prior to the test was 9.4 per cent compared with 3.9 per cent for those who had been scanned (P < 0.0005). Seventy-five per cent (36/48) of Down's syndrome pregnancies were detected for a false-positive rate of 4.0 per cent. Twenty-five out of 36 of detected Down's syndrome pregnancies were dated by scan prior to sampling, and in the 11 remaining cases, the dates were confirmed by scan after a high-risk result was obtained. The exclusion of uE3 from the screening protocol would have reduced the detection rate to 52 per cent (25/48) for the same false-positive rate. Eighty-five per cent of women identified at high risk accepted the offer of an amniocentesis. Other fetal abnormalities detected were trisomy 18 (3), trisomy 13 (2), 45,X (6), 69,XXX (5), other chromosome abnormalities (9), open neural tube defects (26), hydrocephalus (7), abdominal wall defects (4), and steroid sulphatase deficiency (6).