The present study investigates collaborative learners' adoption of key regulation activities (i.e., orienting, planning, monitoring, and evaluating) and a deep-level regulation approach in relation to characteristics of their collaboration on the cognitive and communicative level. More specifically, the correlation of collaborative learners' regulation behavior with respectively their content processing strategies and the level of transactivity in their discussions is analyzed. The study is conducted in a naturalistic reciprocal peer tutoring (RPT) setting in higher education. Sessions of five randomly selected RPT groups participating in a semester-long RPT intervention were videotaped (70 h). Binary logistic regressions were performed to examine how RPT participants' metacognitive regulation is related to their content processing and transactive discussions. Results reveal that students' adoption of key regulation activities is significantly correlated with their adoption of content processing strategies, although different correlations are revealed for particular regulation activities. Additionally, RPT participants' adoption of regulation activities is significantly related to students' transactive discussions, both when reacting to each other's cognitive and metacognitive contributions. With regard to RPT participants' adoption of a deep-level regulation approach, the results show significant correlations with higher-order content processing as well as with representational and operational transactive discussions, in which students respectively paraphrase or elaborate on each other's contributions. The present study's micro-analytical examination of RPT participants' learning and regulation processes contributes important insights to the literature on collaborative learners' regulation, providing input for stronger theoretical models and facilitating instructors' adequate support of collaborative learners.