The paper presents an analysis of semantic and pragmatic features of the double modal ‘might could’ which proves to be the most common of combinations of modal auxiliaries observed in modern English. Although at first glance combinations of two modal verbs might seem to be cases of grammatical distortion, their use is scientifically recognised. The first instances of double modals date back to the seventeenth century. While previous research has focused primarily on formal, pragmatic and sociolinguistic aspects of double modality, this study aims to describe the functional potential of double modal combinations by looking into the semantics of their components. The distinction between epistemic, deontic and dynamic modalities forms the basis for the analysis. It is generally assumed that in the case of ‘might could’ the first-tier component is normally used in the epistemic meaning of supposition implying uncertainty, whereas ‘could’ is either deontic or dynamic. Close analysis of situations where ‘might could’ was used by native speakers of English enabled us to conclude that although the epistemic value of ‘might’ prevailed in the majority of the analysed examples, in about one third of them its use was not purely epistemic. With the deontic ‘could’ the first-tier ‘might’ was clearly a politeness marker, whereas in the case of the dynamic ‘could’ it manifested possible lack of commitment. While the double modal is not recognised as standard British or American English, it does exist as a dialectal feature in both Britain and the US, and this article discusses the use of double modals as a popular dialectal variety of English used in regions of the US and Great Britain.