This study examines the discourse functions of BE expected to and BE supposed to in the genre of judicial opinion, providing insights into discipline-specific practices of epistemological positioning. Drawing on the 130 million words Corpus of US Supreme Court Opinions, it looks at how the two mindsay constructions were deployed in judicial writing over a period of more than 200 years, and identifies divergent frequency patterns associated with their use. As the findings reveal, in the opinions, on the one hand, BE expected to tends to co-occur with reasonably (can/could (not) reasonably be expected to) and is used to create a semblance of objectivity. BE supposed to, on the other hand, favors the present tense and third-person reference (which/it is supposed to) and serves as a distancing device. The paper also compares the frequency patterns involving BE expected to and BE supposed to found in the opinions with those attested in the Corpus of Historical American English, and it demonstrates that judicial writing exhibits trends which clearly differ from trends noted in non-judicial registers.