This study examined the relationship of child gender and maternal race to the looking and talking interactions of fifty-four 3-year-old prematurely born children and their mothers. More gender differences occurred for looking than for talking, whereas racial differences were stronger for talking than for looking. Transitional probabilities between looking situations suggested that gender differences occurred because girls were more likely to respond when their mothers were looking at them than were boys. Transitional probabilities between talking situations suggested that racial differences occurred because non-White mothers (African Americans and Native Americans in this study) were less likely to respond when their children were talking to them than were White mothers. The only significant interaction of gender by race was with White mothers who were more likely to respond when their girls were talking. When analyses were repeated, controlling for the effects of intelligence quotient, gestational age, neurobiologic risk score, and socioeconomic status (SES), gender differences for looking situations became smaller, whereas racial differences for talking situations became larger. Gender and ethnicity differences for looking and talking interactions cannot be explained by simple differences in health status or SES. These differences may possibly be related to the differential brain functions and hormonal effects of boys and girls, as well as to differential socialization that influences gender identity and gender roles. They also may be related to the higher incidence of language delays in non-White children and to differences in sociocultural norms and parenting between White and non-White mothers.