Introduction of maternal serum markers for prenatal screening of Down's syndrome implies a redefinition of the criteria used for identifying at risk women for screening by amniocentesis, which are currently based on maternal age. On the basis of the first French prospective study of human chorionic gonadotropin measurement in maternal serum as a predictor of Down's syndrome, this paper shows that a screening policy combining maternal age with hCG measurement was more cost-effective than one relying on maternal age alone. A cost-benefit analysis (using the <<avoided>> lifelong costs of care for a trisomic 21 child that are allowed by prenatal screening) would justify lower hCG cut-off values and an higher detection rate of Down's syndrome (74,5 %) than other decision rules based on alternative principles such as: equalization to 1 % of fetal risk of Down's syndome for access to amniocentesis at all maternal ages or equalization to 1 parts per thousand of fetal risk of Down's syndrome among women not undergoing amniocentesis. However, results of the cost-benefit analysis are very sensitive to other factors such as the cost associated with loss of a normal fetus due to iatrogenic risk of amniocentesis (false positives of hCG). Ethical and value-ladden issues that necessarily underlie economic evaluation of screening programmes, as well as other decision rules based on equalization of acceptable risk for each maternal age, are discussed.