Massive Open Online Courses (MOOCs) in higher education have received a great deal of attention during recent years (Karsenti, 2013). Udacity, Coursera and EdX, the main providers of MOOCs in the US, are adding universities as partners at a breath-taking speed; the same is true, although to a lesser extent, for MOOC providers in Europe. OpenupEd, for instance, a pan-European initiative founded in 2013 and supported by the European Commission is offering courses from a number of European and even non-European higher education institutions. Also, a number of national institutions in Europe have started to offer MOOCs (see the European MOOCs scoreboard). Despite public enthusiasm concerning MOOCs, participants in MOOCs seem to meet with serious problems leading to enormous dropout rates. A recent study showed that only 4 % of students attending Coursera MOOCs completed their courses (Armstrong, 2014). One of the problems may be that many courses were created without taking into consideration findings of research in the fields of learning and self-regulated learning. The present contribution will therefore suggest criteria derived from theories of learning (with special consideration of connectivism, Siemens, 2005) and self-regulation which should be met in order to support self-regulated learning with MOOCs.