Empirical studies reveal a close relationship between epistemological beliefs (EBs) and metacognition. For example, more "sophisticated" beliefs are associated with more self-reported monitoring strategies. This relationship is also advocated theoretically. Nevertheless, exactly "how" and "why" EBs impact learning is still an open question. In accordance with others (for example Muis 2007; Muis and Franco 2009) we conceive the COPES Model (Winne and Hadwin 1998) as an appropriate answer to the "how" question. Within that model, epistemological beliefs are conceptualized as "internal conditions of learning" and they are embedded into recursive processes of self-regulation. With regard to the "why" question, we assume that EBs function as general ideas about knowledge for the apprehension of the content to be learnt. Such apprehension allows for the calibration of learning to different learning tasks. We review two clusters of studies on the preparatory and the enactment stages of learning testing this apprehension and calibration hypothesis.