Prospective studies of the effects of prenatal alcohol exposure on development have focused primarily on the detection of subtle deficits. This study was designed to extend those findings by evaluating dose-response, functional significance, and pattern of drinking in greater detail. 480 African-American infants, recruited to overrepresent prenatal exposure at moderate-to-heavy levels, were assessed. For the five outcomes tested, nonparametric and hockey stick regression analyses both indicated essentially no relation between pregnancy drinking and developmental outcome below a median threshold of 0.5 oz absolute alcohol/day, with the impact of the exposure increasing gradually above threshold. Functional deficit was defined in terms of performance in the bottom 10th percentile of the distribution. For four of the five outcomes tested, there was no increased incidence of functionally significant deficit in infants born to moderate-to-heavy drinking mothers <30 years old, whereas those born to older drinking mothers were 2 to 5 times more likely to be functionally impaired. Among the infants exposed above threshold, functionally significant deficits were seen primarily in those whose mothers averaged at least 5 drinks/occasion on an average of at least once/week. By contrast, a history of alcohol abuse was not related to functional deficit. These data suggest that efforts to reduce the incidence of alcohol-related functional impairment should specifically target the older mother who engages in intermittent heavy drinking during pregnancy.