Numbers may seem “easy” to interpret, for having fixed correspondences between source and target language. However they are also less predictable, less informationally redundant, and less amenable to terse expression, than other types of information and so are more taxing on working memory, a key cognitive constraint during simultaneous interpreting. The appearance of unpredictable numerical information in a speech causes an increase in cognitive load for the interpreter, which can cause rendering problems. Misalignments between source and target language number systems can add another layer of difficulty. I argue that advice such as note-taking, segmentation, and reformulating the order of information can help mitigate these difficulties to some extent. But numerical information will always be one of the more difficult types of information to interpret, for student and expert interpreters alike. Expert interpreters may be “better at numbers” because of general language proficiency advantages, and because they have acquired long term memory schemas for handling formulaic, non-numerical, parts of speeches, which should also inform advice given to students.