Transitive inference (TI) occurs when information about known relationships is used to construct novel associations, for example, when an individual infers the order of 3 conspecifics in a linear dominance hierarchy (A > B > C) by watching 2 dyads (A > B and B > C). TI has been demonstrated for several species, but never for organisms with hierarchies involving many individuals whose rank order changes often. Under these circumstances, theory predicts natural selection may favor use of explicit cues to rank conspecifics, rather than TI. Here we show that brook trout (Salvelinus fontinalis), a species that exhibits complex, shifting dominance hierarchies, can combine individual recognition and TI to rank individuals within a hierarchy without use of explicit cues. Subjects that were allowed to directly interact with rivals were able to identify the social rank of each individual. In addition, we found that a subject, whose rank was C within a 5-fish hierarchy (A > B > C > D > E), could use TI to correctly infer the ranks of A and E after first interacting with B and D and then watching dyads involving A > B and D > E. This demonstration of TI in brook trout contributes to the understanding of stream trout population dynamics as TI could lower the energetic costs associated with movement.