Anthropometric data and computerized electroencaphalogram analysis during quiet, indeterminate, and active sleep were obtained from infants of mothers of four groups: (1) heavy drinking mothers (greater than 2 ounces of alcohol per day); (2) nondrinking mothers; (3) smoking, nondrinking mothers; (4) nonsmoking, nondrinking mothers. Infants in groups 1 and 2 were matched as closely as possible for postconceptional age, sex, race, and socioeconomic status. Infants in groups 3 and 4 were matched similarly. Infants of alcoholic mothers had a significantly lower birth weight, length, and head circumference than those from the matched control group. Infants of smoking mothers had lower birth weights and lengths than infants of nonsmoking mothers, but head circumference was identical. Hypersynchrony of the electroencephalogram was seen only in "alcoholic" infants, and power spectral density analysis revealed that the average integrated power was significantly increased in quiet, active, and indeterminate sleep. The greatest increase in electroencephalogram power (212%) was seen in active sleep, and this analysis clearly separated 15 of 17 alcohol-exposed infants from the control infants. These data suggest that alcohol has a specific toxic effect on the fetal brain that is not linked with smoking habits. The neonatal electroencephalogram is affected even in the absence of dysmorphology and thus may be the most sensitive indicator of fetal alcohol toxicity.