This paper describes the process of acquisition in learner varieties with respect to reference and referential movement in the domain of modality. The findings are based on data from the longitudinal ESF and P-MoLL projects and on cross-sectional data of Italian learners of German, as well as German and Italian native speech. The theoretical framework is provided by Klein and von Stutterheim's (e.g., 1987) “quaestio model” and their concept of referential movement. The concept of modality is based on Dietrich's (1992) theory of modality. The present findings show that, in instructional discourse, the German native speakers prefer implicit, contextual-based modal means when referring to maintained topic information in the domain of modality, whereas in the learner varieties at least three main stages can be observed: a phase of formulaic speech and pragmatic mode, a phase of high explicitness, and a phase of approach toward implicit reference based on (target) principles of referential movement. With the help of a new category—subquaestio—I show how, with respect to the change of modal means, the use of explicit modal marking in German native speech generally arises from specific local difficulties. In contrast, the use of explicit modal reference in learner varieties remains to a large extent unaffected by whether the modal marking depends on the overall text quaestio or on local problems.