We investigated the relationship between maternal thyroid disease and the risk of birth defects in offspring using data from a large population-based, case-control study. Cases included 4904 stillborn and liveborn infants with major anomalies diagnosed in the first year of life and born to residents of metropolitan Atlanta between 1968 and 1980. Controls included 3027 normal babies, frequency-matched to cases by race, hospital of birth and quarter of birth. We compared mothers of cases and controls regarding history of physician-diagnosed hypothyroidism and hyperthyroidism before the infant's birth, age at diagnosis of thyroid condition, duration of illness, and intake of thyroid medications before and during pregnancy. Information obtained from maternal interviews was evaluated for concordance with hospital records. We adjusted for potentially confounding factors using conditional logistic regression analysis. Overall, there was no relationship between the risk of total birth defects and history of maternal hypothyroidism (odds ratio (OR) = 1.05, 95% C.I. 0.84-1.31), maternal hyperthyroidism (OR = 1.00, 95% C.I. 0.66-1.53), and intake of thyroid hormone and antithyroid drugs before and during pregnancy. In an analysis of 66 specific birth defects and defect groups, we found two statistically significant associations with hypothyroidism and three with hyperthyroidism which may reflect chance findings. In an evaluation of babies with multiple anomalies, we observed a two-fold increased risk with hypothyroidism but no discernible pattern of defects. The absolute risk of major birth defects in offspring of women with history of hypothyroidism can be estimated as 2.1%, a finding at odds with the 10-20% risk cited in the literature.