The risk of an individual woman having a pregnancy associated with Down's syndrome was estimated from her age and her serum alpha-fetoprotein (AFP) level at 14-20 weeks gestation. The estimates were made using published data on the risk of Down's syndrome in relation to maternal age from 4528 affected and over 5 million unaffected pregnancies, and on the risk in relation to serum AFP from 68 affected and 36,645 unaffected pregnancies. Separate estimates were derived for AFP levels using gestational age based on the time since the first day of the last menstrual period and on an ultrasound biparietal diameter measurement. In each case this was done with and without adjusting AFP levels to take account of maternal weight. The same sources of data were also used to construct six Down's syndrome screening policies, each combining information on maternal age and serum AFP. For example with one policy the detection rate would be 28% and would involve selecting 2.8% of unaffected pregnancies for amniocentesis; using age alone the same detection rate could only be achieved by selecting 4.3% of unaffected pregnancies for amniocentesis--an increase of 50%. In general, screening for Down's syndrome using both maternal age and serum AFP is more efficient than either alone.