Soil carbon (C) losses and soil translocation from tillage operations have been identified as causes of soil degradation and soil erosion. The objective of this work was to quantify the variability in tillage-induced carbon dioxide (CO₂) loss by moldboard (MP) and chisel (CP) plowing across an eroded landscape and relate the C loss to soil properties. The study site was a 4ha wheat (Triticum aestivum L. cv. Marshall) field with rolling topography and five soil types in the Svea-Barnes complex in west central Minnesota (N. Latitude=45°41′W, Longitude=95°43′). Soil properties were measured at several depths at a 10m spacing along north–south (N–S) and west–east (W–E) transects through severely eroded, moderately eroded and non-eroded sites. Conventional MP (25cm deep) and CP (15cm deep) equipment were used along the pre-marked transects. Gas exchange measurements were obtained with a large, portable chamber within 2m of each sample site following tillage. The measured CO₂ fluxes were largest with the MP>CP>not tilled (before tillage). The variation in 24h cumulative CO₂ flux from MP was nearly 3-fold on the N–S transect and 4-fold on the W–E transect. The surface soil organic C on the transects was lowest on the eroded knolls at 5.1gCkg⁻¹ and increased to 19.6gCkg⁻¹ in the depositional areas. The lowest CO₂ fluxes were measured from severely eroded sites which indicated that the variation in CO₂ loss was partially reflected by the degradation of soil properties caused by historic tillage-induced soil translocation with some wind and water erosion. The spatial variation across the rolling landscape complicates the determination of non-point sources of soil C loss and suggests the need for improved conservation tillage methods to maintain soil and air quality in agricultural production systems.