The authors performed surveillance for fetal alcohol syndrome with an existing birth defects registry. Fetal alcohol syndrome cases were identified from multiple sources using passive surveillance and from two selected medical sites using enhanced surveillance. Between May 1992 and March 1994, a total of 173 cases were identified, and the medical records of the cases were reviewed to determine whether the cases met a surveillance case definition for fetal alcohol syndrome. Of these cases, 37 (21 percent) met either definite (28) or probable (9) criteria for fetal alcohol syndrome, 76 met possible criteria (44 percent), and 60 (35 percent) were defined as not fetal alcohol syndrome. Enhanced surveillance had the highest sensitivity for definite or probable cases, 31 of 37 (84 percent), followed by hospital discharge data, 14 of 37 (38 percent). The authors also compared birth certificate information for 22 definite or probable cases in children born between 1989 and 1992 to birth certificate information for all Colorado births for that period. The proportion of mothers of children with fetal alcohol syndrome was statistically significantly greater (as determined by exact binomial 95 percent confidence limits) than the proportion of all mothers for the following characteristics: black race (0.23 versus 0.05), unmarried (0.55 versus 0.22), not employed during pregnancy (0.86 versus 0.43), and started prenatal care in the third trimester (0.18 versus 0.04). Surveillance for fetal alcohol syndrome can be accomplished with an existing registry system in combination with additional case finding and verification activities. Through followup investigation of reported cases, data can be gathered on the mothers of children with fetal alcohol syndrome. These data could be used to target fetal alcohol syndrome prevention programs.