The ability to mindread recursively-for example, by thinking what person 1 thinks person 2 thinks person 3 thinks-is a prime example of recursive thinking in which one process, representation, or idea becomes embedded within a similar one. It has also been suggested that mindreading is an exceptional example, with five recursive steps commonly observed for mindreading, in comparison with just one or two in other domains. However, conceptual analysis of existing recursive mindreading tasks suggests that conclusions about exceptional mindreading are insecure. Revised tasks were devised to provide a more rigorous test of recursive mindreading capacity. Study 1 (N = 76) found significantly worse performance at level-5 recursive mindreading on the revised tasks (17% correct) compared with the original tasks (80% correct), and no effect of moderate financial bonuses for good performance. Study 2 (N = 74) replicated poor performance at level-5 recursive mindreading on the revised tasks (15% correct) in the absence of bonuses, but found better performance (45% correct) when participants were offered large bonuses for accuracy, encouraged to take as much time as needed, and assisted with a strategy for recursive reasoning. These findings suggest that, like recursive thinking in other domains, recursive mindreading is effortful and limited. We discuss how the proposed role for high levels of recursive mindreading in communication, culture, and literature might be reconciled with these limitations. (PsycInfo Database Record (c) 2023 APA, all rights reserved).