The Sucker Creek Formation exposures at Devils Gate in southeastern Oregon have yielded a significant small mammal fauna of at least thirty small mammal taxa from five stratigraphic horizons. The mammal-bearing portion of the Devils Gate section is more than 200 m thick. Fossil mammals occur in lacustrine and marginal lacustrine deposits lower in the section and occur in overbank and paleosol deposits higher in the section. ⁴⁰Ar/³⁹Ar single-crystal laser-fusion dates on three Devils Gate ashes shows that the age of the mammal-bearing sequence at Devils Gate spans the late early Barstovian land-mammal age with possible overlap into the late Barstovian, as currently defined. Duration of the entire mammal-bearing portion of the Devils Gate section was less than a million years. Both a new ash date from the type section and biostratigraphic correlations between Devils Gate and the type section support considerable temporal overlap between the two exposures. The Devils Gate Local Fauna includes several new taxa: a phyllostomatid bat; two "flying squirrels", Petauristodon sp. A and Petauristodon sp. B; and an eomyid rodent, Leptodontomys sp. A. Several fossil occurrences represent the first record of a taxon in the northern Great Basin and/or in the Barstovian land-mammal age, including: Blackia sp., Schaubeaumys grangeri, Protospermophilus quatalensis, and Pseudadjidaumo stirtoni. The Stagestop locality produced two new taxa, Copemys sp. aff C. esmeraldensis and Mystipterus sp. The Stagestop local fauna is Clarendonian in age. Concretions are an important source of fossil mammals in exposures of the Sucker Creek Formation. Geochemical analyses show that concretions formed through a complex interaction between bone and surrounding volcaniclastic material. Although some superficial bone was consumed during concretion diagenesis, concretion development reduced the chance of prolonged chemical and physical destruction of bone during later soil development. The broad ecological diversity of small mammals recovered from Devils Gate supports an interpretation of the local paleoecology as a mosaic of grassland, forest, and pond/lake-bank environments. Sequential small mammal faunas across a prominent ash event show a generally stable composition with no pronounced ecomorphic differences in pre- and post-volcanic disturbance intervals. Therefore, small mammals do not show analogous ecological patterns to disturbance-driven plant successions in the Sucker Creek Formation. I infer that the local ecosystem recovered from volcanic blasts at a temporal scale below the resolution of time-averaged, post-disturbance paleosols.