Mutual regulation of attention was investigated in a group of prenatally cocaine-exposed and non-exposed mother-infant dyads during a 5-minute videotaped free play session. Mutual regulation was measured using a state-based coding scheme designed to categorize dyadic interactions into three mutually exclusive and exhaustive states: maternal bid, mutual engagement, and non-involved. Results revealed no significant differences between cocaine-exposed and non-exposed dyads in overall amount of mutual engagement displayed. Cocaine-exposed dyads exhibited significantly longer mutual engagement episodes. Mothers in the two groups did not differ in the number or quality of bids for mutual engagement, and infants in both groups were equally responsive to maternal bids. No ecological variables were found to predict mutual engagement.