The ability to display caring responses to another child's distress is a key aspect of early empathy that is facilitated by parental socialization. However, existing studies typically involve lab settings and focus on toddlers' unsupported responses to adult simulations of distress, raising questions about their ecological validity. Framed within the New Fathers and Mothers Study, the current study involved 156 British toddlers (Mage = 24.35 months, SD = 0.73) who were filmed at home with either their mother or father (87 mothers and 69 fathers) in a novel paradigm involving a lifelike crying baby doll. Capitalizing on the inclusion of both fathers and mothers, a key question concerned effects of parent-toddler dyad gender composition on both global ratings of toddlers' displays of empathic concern and more specific indicators, including toddlers' attentional, emotional, and behavioral responses. Whereas parental responses did not differ by either child or parent gender and appeared to be closely attuned to child behavior, toddlers' responses showed effects of both (a) child gender, evident in higher rates of emotion labeling in girls than in boys (even when controlling for language ability), and (b) parent gender, evident in higher levels of empathic concern for girls observed with fathers than for those observed with mothers. These findings are discussed within the context of empathy development and parental socialization.