Background: Home visiting programs constitute an important policy to support vulnerable families with young children. They mainly aim to improve infant-parent relationships, however evidence on their effectiveness based on observational measures is relatively scarce. The present study provides the rare opportunity to directly examine the effects of a home visiting program, the Pro Kind, on mother-child interactions in a randomized controlled trial (RCT). Methods: A sample of 109 mother-child dyads was videotaped during a 3-min typical play situation at the participants' homes when the child was aged 25 months. We use a novel micro-coding system which allows us to examine how the intervention affected the dynamic feedback responses of both mothers and children in three key measures of behavior: orientation, positive contingency, and negative/lack of contingency. The study is registered in the German Clinical Trial Register (trial ID: DRKS00007554, date of registration: 11 June 2015). Results: The intervention significantly improved the interactions between girls and their mothers, both in strongly stable and partly unstable situations. Mixed impacts were detected for boys. Conclusions: These results have important implications for the analysis of mother-child interactions data and the design of home visiting programs.