OBJECTIVE: To evaluate the impact of policy and practice changes in prenatal screening for Down's syndrome on prenatal diagnosis and live birth prevalence of Down's syndrome. DESIGN: Population-based observational study. SETTING: Greater Paris. POPULATION: Residents of Greater Paris who gave birth or had a termination of pregnancy in Paris during 1981-2000 (approximately 38,000 births per year). METHODS: Data on 1916 cases of Down's syndrome were obtained from the Paris Registry of Congenital Anomalies. Analyses included binomial and Poisson models of trends in three periods: prior to 1989 (reference period), 1989-1995 (reimbursement of amniocentesis in case of ultrasonographic anomalies) and 1996-2000(widespread use of reimbursed serum screening and measurement of nuchal translucency). MAIN OUTCOME MEASURES: Trends in proportion of Down's syndrome cases diagnosed prior to birth; live birth prevalence of Down's syndrome. RESULTS: The proportion of Down's syndrome detected prenatally for women <38 years of age increased ninefold; from 9.5% (95% CI 2.7-22.6) in 1981 to 84.9% (95% CI 74.6-92.2) in 2000. For women >38 years of age, the increase was 1.5-fold. The live birth prevalence of Down's syndrome decreased by 3% per year (prevalence ratio [PR] 0.97, 95% CI 0.96-0.99); the age-adjusted decrease was 13%. The analysis by period showed that the decrease in live birth prevalence of Down's syndrome was greater after 1988. CONCLUSIONS: By far, most cases of Down's syndrome are currently detected prenatally in the Parisian population. Consequently, the live birth prevalence of Down's syndrome has decreased despite consistent trends towards delayed childbearing. These positive public health effects have to be balanced against a relatively high rate of amniocentesis and the potentially negative consequences of widespread prenatal testing for individuals born with Down's syndrome.