Licensed hospitals usually maintain a discharge diagnoses index, which provides an inexpensive tool for the surveillance of birth defects diagnosed shortly after birth. Government agencies in several states routinely use discharge diagnoses for this purpose. To evaluate the completeness of the discharge diagnoses, the authors compared birth defects noted in the discharge diagnoses with those noted anywhere in the hospital birth record in a cohort of 3,421 infants born to US Army veterans from 1966 to 1986. In this cohort, 237 birth defect cases were documented in hospital birth records, and 49% of those cases were missed in the discharge diagnoses (28% of major defect cases and 66% of minor defect cases). The extent of missed defects varied greatly by organ system and by specific defect. Significant predictors of a missed defect were the presence of multiple defects, female sex, and western region of birth. The underascertainment of defects in the discharge diagnoses should be considered in the development and operation of surveillance systems using this source of data.