Although the excess risk for birth defects among children of mothers with diabetes mellitus is well documented, there are few data concerning the risk for specific malformations. In the Atlanta Birth Defects Case-Control Study, those risks for malformations were evaluated. The population-based study included 4929 live and stillborn babies with major malformations ascertained by the Metropolitan Atlanta Congenital Defects Program in the first year of life born to residents of Metropolitan Atlanta between 1968 and 1980. The study also included 3029 nonmalformed live babies who were frequency-matched to case babies by race, period of birth, and hospital of birth. The relative risk for major malformations among infants of mothers with insulin-dependent diabetes mellitus (n = 28) was 7.9 (95% confidence interval [CI]1.9, 33.5) compared with infants of nondiabetic mothers. The relative risks for major central nervous system and cardiovascular system defects were 15.5 (95% CI = 3.3, 73.8) and 18.0 (95% CI = 3.9, 82.5), respectively. The absolute risks for major, central nervous system, and cardiovascular system malformations among infants of diabetic mothers were 18.4, 5.3, and 8.5 per 100 live births, respectively. Infants of mothers with gestational diabetes mellitus who required insulin during the third trimester of pregnancy were 20.6 (95% CI = 2.5, 168.5) times more likely to have major cardiovascular system defects than infants of nondiabetic mothers. The absolute risk for infants of this group of diabetic mothers was 9.7%. No statistically significant differences were found among infants of mothers with gestational diabetes mellitus who did not require insulin during pregnancy.(ABSTRACT TRUNCATED AT 250 WORDS)