Why do humans implicitly categorize individuals by their race? The alliance hypothesis-which argues that racial categorization is a byproduct of evolved information-processing systems in the mind for keeping track of alliances-has been the most successful causal account of racial categorization to date, amassing a large number of uniformly successful findings across a number of different papers. These findings show that when race is crossed with (i.e., is not predictive of) alliance membership, participants' categorization by race is reduced or eliminated, whereas categorization by other dimensions, such as sex or age, remain relatively unaffected. These results have been taken to mean that race is an alliance proxy within the mind, and can be superseded when better alliance information is provided. However, a counterhypothesis remains that cannot only account for all of the data observed in this past work, but could also undermine the entire theoretical interpretation of those data, which is that race may simply be a more flexible social category-meaning that categorization by race will be lowered by any crossed category, even categories that are not alliances. This counterhypothesis has never been tested against. The study reported here does so, examining if a contextually relevant crossed category that is not an alliance also reduces racial categorization. Results demonstrate that it does not: neither race nor sex (between-subjects) was affected when crossed with this nonalliance category. The race-is-just-more-flexible counterhypothesis is excluded as alternative causal account of the alliance hypothesis findings on racial categorization. (PsycInfo Database Record (c) 2022 APA, all rights reserved).