We reviewed our experience with a policy of revising the estimated probability of an open neural tube defect for those cases in which maternal serum alpha-fetoprotein (MSAFP) was elevated and the ultrasound examination was normal. Assuming a sensitivity for the ultrasound diagnosis of neural tube defects of 90%, the risk estimate derived from the MSAFP was reduced by that amount when ultrasound was adequate and normal. When counseled regarding this revised risk estimate, 67% of our patients declined amniocentesis. The sensitivity for the ultrasound diagnosis of anencephaly in our series was 100%; for open spine defects, 80%; and for all open neural tube defects, 91%. The two patients in whom an open spine defect was not detected with ultrasound had elevated revised risk estimates. When appropriate criteria are met, the estimated sensitivity of ultrasound in diagnosing open neural tube defects may be used for counseling patients at risk for a fetus with an open neural tube defect.