The interpretation of empty elements, i.e. signs that have no phonetic realization, constitutes a "classical" problem in modern linguistics. Null elements have been postulated on various levels of the linguistic system and its pragmatic use, in particular, in morphology and syntax. An adequate theory of discourse comprehension also requires an account of what is not said but only conversationally implicated. In the last thirty years the interpretation of empty subjects in non-finite clauses, which is known as the "control problem", has attracted the attention of many formal syntacticians. It has however become increasingly clear that the interpretation of such empty elements is only minimally guided by syntactic principles; in addition, a number of semantic and pragmatic factors have to be taken into consideration. The aim of our contribution is to sketch a cognitively based theory of "obligatory" control that explains how general control principles interact with language-specific coding devices. We focus on German data; we surmise, however, that they also hold for other languages. In particular, we aim at elucidating the interplay of syntactic, semantic and pragmatic information in the comprehension of empty subjects in non-finite complement clauses. Using German and, to a lesser extent, English examples, we will demonstrate that the control principles we postulate account for numerous control verbs and control verb classes.