This article is a diachronic study of different constructions involving the verb DARE from Old English (OE) to Modern English (ModE). With regard to the Late ModE (lModE) change, this paper examines the Corpus of Historical American English (COHA) covering 1810 to 2009. It is found that the general tendency is a decline of the frequency of DARE in terms of both the modal auxiliary and the main (lexical) verb uses in American English. However, the frequency of the auxiliary use, contra Taeymans (2004), has more rapidly decreased, while that of the main verb use has less drastically dropped. The blend constructions have had a low frequency throughout the past 200 years. Yet, the inflected blends(e.g. dared, dares) showed gradual decrease of frequency, whereas the DO-support blends did not. This study also examines historical changes from OE to Early ModE (eModE). It is found that lexical and modal properties coexisted from OE to Early Middle English (eME), the modal ones climaxing during the eME period. This paper argues that the grammaticalization from lexical to modal was triggered before the OE period and thereafter until eME is the period of gradual extension. From the lME period lexical properties increasingly occurred, which led Beths (1999) and Taeymans (2004) to argue for degrammaticalization. This paper accounts for the same phenomenon in terms of renewal, a natural process in grammaticalization. There are lME evidences showing phonological and semantic reductions of DARE, which necessitated a renewal of DARE. This study, moreover, shows that the words of the same category can follow different paths of change with time.