Tropical land-use change is the key driver of the global extinction crisis. Understanding of the effects of land-use change on biodiversity depends on using a method that can accurately detect impacts. Studies evaluating conservation questions using small mammals in the tropics are mostly conducted using live-trap or pitfall, yet those that use both methods revealed substantial variation in the range and number of species sampled. A key question is whether variation between trapping methods alters the interpretation of the impacts of land-use change on small mammals. Focusing on the fragmented Brazilian Atlantic Forest, we sampled across five landscape units: interior and edge of contiguous forest; riparian forest remnant connected near or far from the contiguous forest; and unconnected riparian remnant. We found contrasting results from live-trap and pitfall regarding species composition, community structure, and functional abundance of small mammals. We then show that interpretations of land-use change impacts on small mammal communities differ significantly when using live-trap versus pitfalls. Using live-traps, riparian forest closely connected to contiguous forest supported similar composition and structure as the contiguous forest, but similarity decreased with distance from contiguous forest sinks. With pitfalls, the estimated small mammal community did not differ among habitats. Methodological variation could result in key impacts of land-use change on small mammals going undetected and in erroneous conservation planning within fragmented landscapes, indicating either a necessity to use live-traps or a combination of both methods.