A naïve reliability coder was trained on ten videos (five dyads) and then rated one video for each of the remaining thirteen dyads, equating to seven baseline recordings and six outcome recordings (36 % of the total). Calculation of inter- rater reliability was carried out on the mean level achieved by the dyad throughout the interaction. Inter-rater reliability was excellent for both Co-Regulation (ICC = 0.86) and Intersubjective Engagement (ICC = 0.91).