Joint attention between infants and mothers is known to be a predictive factor for individual differences in language and social development of the child. Approaches to measuring joint attention, however, had limitations in securing both objective and naturally valid measures. The present study combined prior methods to obtain naturally valid and systematically comparable joint attention measures in both toddlers and their mothers. We presented semi-structured play setting that can utilize Early Social Communicative Scale and coded entire 30-minute mother-child interactions for initiation of joint attention (IJA) behaviors. Overall, mothers showed more IJA than their toddlers. Mothers with girls tended to use pointing with gaze alternation than mothers with boys. Furthermore, maternal education and empathic accuracy were positively associated with their toddlers’ IJA. However, mothers’ personal distress was negatively associated with toddlers’ IJA. Also, the more the mothers used pointing to initiate, the better they appeared to convey word meanings non-verbally. An association was found between Toddlers’ IJA and their comprehension/production vocabulary and mothers’ IJA was positively linked to their 14-17-month-olds’ word comprehension. These findings suggest possible factors that can facilitate IJA in mother-child interactions.