BACKGROUND: Advances in prenatal diagnosis have led to changes in the management of pregnancies affected with birth defects. These changes pose unique challenges for birth defects monitoring programs which use hospital-based sources. METHODS: In 1994, Metropolitan Atlanta Congenital Defects Program (MACDP) abstractors began to visit area perinatologists' offices to identify pregnancies diagnosed prenatally with fetal defects. These pregnancies were then linked with existing MACDP cases and the hospital deliveries abstracted. Those without a hospital delivery were included as having unknown outcomes. Prenatally diagnosed defects were classified as definite or possible based on the certainty of the prenatal description. For 1995–2004, we calculated minimum and maximum adjusted defect prevalences by adding definite prenatal defects, and definite plus possible prenatal defects, to the hospital-based cases. RESULTS: We identified 1009 pregnancies with a prenatally diagnosed defect not ascertained from MACDP hospital sources. Including these increased the total defect prevalence from 28 per 1000 live births to a minimum of 29.94 (6.9% increase) and maximum of 30.14 (7.7% increase) per 1000. The minimum increase was greater than 50% for conjoined twins, triploidy, craniorachischisis, cystic hygroma, Klinefelter syndrome, anencephaly, Turner syndrome, and trisomies 13, 18 and 21 among mothers ≥35. CONCLUSIONS: These data reflect the variety of congenital abnormalities that can be detected prenatally and the importance of including prenatal diagnoses in birth defects monitoring data. Birth defects monitoring programs should assess individually the extent to which prenatal diagnosis can affect the accuracy and completeness of their data. Birth Defects Research (Part A), 2009. Published 2008 Wiley-Liss, Inc.