In some cooperatively breeding groups, individuals have distinct behavioral characteristics that are often stable and predictable across time. However, in others, as in the eusocial naked mole-rat, evidence for behavioral phenotypes is ambiguous. Here, we study whether the naked mole-rat can be divided into discrete phenotypes and if circulating hormone concentrations underpin these differences. Naked mole-rat colonies consist of a single breeding female and large numbers of non-reproductive subordinates that in some cases can exceed several hundred in a colony. The subordinates can potentially be divided into soldiers, who defend the colony; workers, who maintain it; and dispersers, who want to leave it. We established six colonies de novo, tracked them over three years, and assessed the behavior and hormone concentrations of the subordinates. We found that soldiers tended to be from earlier litters and were higher ranked compared to workers, whereas dispersers were distributed throughout litters and rankings. There was no difference in estradiol, testosterone, or dehydroepiandrosterone (DHEA) concentrations among phenotypes. Progesterone concentrations were higher in soldiers, but this difference appeared to be driven by a few individuals. Principal component analysis demonstrated that soldiers separated into a discrete category relative to workers/dispersers, with the highest ranked loadings being age, body mass, and testosterone concentrations. However, the higher testosterone in soldiers was correlated with large body size instead of strictly behavioral phenotype. Workers and dispersers have more overlap with each other and no hormonal differences. Thus the behavioral variation in subordinate naked mole-rats is likely not driven by circulating steroid hormone concentrations, but rather it may stem from alternative neural and/or neuroendocrine mechanisms.