The English construction “be supposed to X” is used in a variety of functions in Present-day English, including evidential, epistemic, and deontic functions. This research offers description and explanations for the development of the evidential, epistemic, and deontic functions from an earlier passive construction, through distinct processes of reanalysis (Hopper and Traugott 1993). I argue that the motivations for these semantic and syntactic shifts are motivated by pragmatic inferences based on: discourse function, discourse expectations about human subjects, frequency effects related to semantic properties of the construction in discourse, and reader-writer expectations about genre type. The results indicate that the evidential function is not part of the general category of epistemicity for this construction, following de Haan (1999, 2001b); that this construction does not exhibit the predicted pathway of semantic development from deontic to epistemic functions (Traugott 1989) due to constraints imposed by the source construction; and that genre plays an important role not only in the relative frequency of the construction (Biber et al. 1999), but also in the emergence of the deontic function diachronically. Finally, I situate the construction in relation to cross-linguistic patterns (Bybee et al. 1994), noting how it parallels broader patterns in the development of the deontic function.