A prospective study of maternal serum alphafetoprotein (alpha-FP) screening of 9838 women in an area with low prevalence of neural tube defects and predominance of anencephalics revealed that an intervention point of single serum alpha-FP level above 2.8 times the median was appropriate for this population. Ninety per cent of anencephalics and all fetuses with anterior abdominal wall defects were detected. There was no spina bifida among the population screened. Two per cent of the population screened had serum alpha-FP level above this cut-off level. Thirty-two per cent of twin pregnancies, 7 per cent of small-for-gestational age infants and 9 per cent of pregnancies which ended in either abortion or perinatal death in the population screened also had one serum alpha-FP level above this intervention point. The false positive rate was 66 per cent. This false positive rate was only reduced to 63 per cent if instead of one, two serum alpha-FP level above this intervention point was considered abnormal. Using this strategy there was no significant reduction in the detection rate of fetal anomalies and other pregnancy complications. Because of the predominance of anencephalics in this population the diagnosis of fetal anomaly in women with abnormal serum alpha-FP level was made by ultrasound examination alone. The reason amniocentesis was not performed in these patients was to avoid unnecessary loss of normal pregnancies which may result from this procedure.