This study is concerned with the use of the English modals (may, might, can, could, shall, should, will, would and must) in adverbial, relative complement clauses. It employs synchronic data from the British National Corpus and quantitative methods to investigate similarities and differences between the core modals, as well as modal-specific preferences in subordinate clauses. The main finding is that modal verbs in subordinate clauses may be conceived of as meso-constructions and that they qualify as micro-constructions once further syntagmatic features are considered. This allows for distinguishing modal verb phrases with different degrees of complexity, schematicity, productivity and subjectivity. Further applications give us insights into collocations, modal harmony, semantic preference, and the attraction of dynamic meaning to relative clauses.