We describe the theoretical model underlying our interventions in metacognition and self-regulated learning (SRL). Our studies are the only ones in which all SRL components are trained in various learning domains and transfer between these domains is explicitly addressed, and that use both offline and online measures to evaluate the effects of the intervention. Pre-experimental and quasi-experimental studies in regular classes were conducted. Offline and online measures were used to evaluate the effects of the interventions. Self-regulated learning can be enhanced in students with learning difficulties and intellectual disabilities, as well as in typically developing students. Improved SR behaviours are accompanied by improved academic performance. Typical problems associated with data collecting using questionnaires (offline) were encountered. On the other hand, online measures require an important investment of resources. Therefore, the number of participants was limited in the observational studies, which makes generalisation more difficult. Furthermore, it must be acknowledged that quasi-xperimental studies are difficult to realise in a natural context. Interventions based on metacognition and self-regulated learning can be effectively implemented in various natural settings. The results of our first series of studies are very positive, but more studies with control groups and delayed post-test to assess maintenance of the learned skills are needed.