The authors examined relations among observed joint attention, maternal report of child’s social competence, setting (home vs. laboratory), task (unstructured vs. semi‐structured), and dyad type [hearing mother–hearing child (n = 55, Mage = 25.8 months) vs. hearing mother–deaf child (n = 27, Mage = 26.9 months)]. Hearing child dyads scored higher on joint attention during unstructured tasks, especially in their home environment. Hearing child dyads displayed similar joint attention to deaf toddler dyads when they engaged in a semi‐structured task, but higher on these measures during unstructured free play. Unlike hearing children, joint attention was differentially related to social competence in deaf children, with relatively higher versus lower social competence depending on relatively high versus low observed joint attention, respectively.