Use of stable carbon isotope data from paleosols to reconstruct past plant community structure (C 3 vs. C 4) has become commonplace. In an effort to improve our ability to make isotope-based reconstructions and to better appreciate the pitfalls, investigations were conducted on both modern soils and paleosols in the Kansas grasslands. Stable carbon isotope data were derived from soils and vegetation on the near-pristine, C 4-dominated grassland of the Konza Tallgrass Prairie Long-Term Ecological Research (LTER) site in northeastern Kansas. In order to evaluate variation of δ 13C within the landscape, two levels of sampling were employed: 2 m-deep upland cores extracted to assess variation with depth in the soil profile, and, to assess variability across the landscape, surface samples along two transects and from within a 660×690 m grid. For transect and grid points, both the upper 2 cm of sediment and the aboveground biomass were collected. Core samples taken at Konza reveal that soil organic carbon was depleted in 13C within the upper 10–20 cm relative to the remainder of the soil solum below, a phenomenon previously reported. In transects and the sample grid, soil organic carbon from soil surfaces was consistently more depleted in 13C than aboveground tissue of associated vegetation samples. Slope, azimuth, and insolation were computed from field data and a high-resolution DEM of the sample grid, but these variables offered no significant explanation of the spatial variability in δ 13C data from soil organic carbon. The observation that modern landscape position has little effect on δ 13C over short distances, at least in a C 4-dominated community, lends support to the application of δ 13C analysis to buried soils for paleoenvironmental reconstruction where paleolandscape position is often unknown. Stable carbon isotope data were derived from paleosols at from the Beisel-Steinle site located to the west of Konza in central Kansas. Three soils were examined: the interstadial paleosol within the Gilman Canyon Formation (c. 38–27k cal yr BP), the Late Pleistocene–Holocene Brady Soil, and the modern surface soil. With the exception of isotopic depletion at the top of the surface soil, all three soils exhibited similar δ 13C values of about14‰, the same value realized in the soil cores from Konza, suggesting close similarity among the plant communities of the interstade, the Late Pleistocene–Early Holocene transition, and pristine prairie of today. This investigation emphasizes the need for additional research into the variation in stable carbon isotope signals vertically within the profile and across the landscape. Issues include the source of the near-surface depletion zone, the relationship between isotopic values in the soil versus those of the vegetation, and variation in soil δ 13C values δ 13C throughout the landscape.