Research on imperative constructions has revealed that imperatives in many languages exhibit common properties, such as subject omission, resistance to combine with negation and non-embeddability (among others) (cf. Han 2001; Aikhenvald 2010). From a semantic/pragmatic perspective, imperatives are an interesting case regarding the division of labor between semantics and pragmatics, since a sentence like Go home! can convey a variety of meanings (imperative speech acts) such as commands, permissions, advice, etc. (cf. Portner 2007, Kaufmann 2011; Condoravdi and Lauer 2012). First studies on sign language imperatives show that some general properties of imperatives seem to hold across modalities, and that imperatives are marked as a sentence type by manual markers while the imperative speech acts are distinguished by manual and non-manual prosodic elements (cf. Donati et al. 2017; Brentari et al. 2018). In the first part of this talk, I will give a short overview of the properties of imperative constructions in spoken languages and of the formal theories that have been suggested for imperatives. Then, I will present a part of my empirical studies on sign language imperatives that come from controlled elicitation tasks. The new findings on imperatives in German (DGS) and Greek Sign Language (GSL) show that imperative speech acts pattern systematically with specific signs, are distinguished by particular clusters of non-manual elements (and different degrees of intensity) and exhibit similar syntactic properties to the spoken language imperatives.